HyperDbg Documentation
CommunityDownloadResearchTutorialhwdbg
  • HyperDbg
  • Getting Started
    • Quick Start
    • FAQ
    • Build & Install
    • Attach to HyperDbg
      • Attach to a remote machine
      • Attach to local machine
      • Start a new process
      • Attach to a running process
  • Using HyperDbg
    • Prerequisites
      • Operation Modes
      • How to create a condition?
      • How to create an action?
      • Signatures
    • User-mode Debugging
      • Principles
      • Examples
        • basics
        • events
          • Getting Results of a System-call
    • Kernel-mode Debugging
      • Principles
      • Examples
        • beginning
          • Connecting To HyperDbg
          • Configuring Symbol Server/Path
        • basics
          • Setting Breakpoints & Stepping Instructions
          • Displaying & Editing & Searching Memory
          • Showing & Modifying Registers and Flags
          • Switching to a Specific Process or Thread
          • Mapping Data & Create Structures, and Enums From Symbols
        • events
          • Managing Events
          • Hooking Any Function
          • Intercepting All SYSCALLs
          • Monitoring Accesses To Structures
          • Triggering Special Instructions
          • Identifying System Behavior
        • Scripting Language Examples
    • Software Development Kit (SDK)
      • Events
        • Conditions
        • Actions
      • IOCTL
        • Event Registration
  • Commands
    • Debugging Commands
      • ? (evaluate and execute expressions and scripts in debuggee)
      • ~ (display and change the current operating core)
      • a (assemble virtual address)
      • load (load the kernel modules)
      • unload (unload the kernel modules)
      • status (show the debuggee status)
      • events (show and modify active/disabled events)
      • p (step-over)
      • t (step-in)
      • i (instrumentation step-in)
      • gu (step-out or go up)
      • r (read or modify registers)
      • bp (set breakpoint)
      • bl (list breakpoints)
      • be (enable breakpoints)
      • bd (disable breakpoints)
      • bc (clear and remove breakpoints)
      • g (continue debuggee or processing kernel packets)
      • x (examine symbols and find functions and variables address)
      • db, dc, dd, dq (read virtual memory)
      • eb, ed, eq (edit virtual memory)
      • sb, sd, sq (search virtual memory)
      • u, u64, u2, u32 (disassemble virtual address)
      • k, kd, kq (display stack backtrace)
      • dt (display and map virtual memory to structures)
      • struct (make structures, enums, data types from symbols)
      • sleep (wait for specific time in the .script command)
      • pause (break to the debugger and pause processing kernel packets)
      • print (evaluate and print expression in debuggee)
      • lm (view loaded modules)
      • cpu (check cpu supported technologies)
      • rdmsr (read model-specific register)
      • wrmsr (write model-specific register)
      • flush (remove pending kernel buffers and messages)
      • prealloc (reserve pre-allocated pools)
      • preactivate (pre-activate special functionalities)
      • output (create output source for event forwarding)
      • test (test functionalities)
      • settings (configures different options and preferences)
      • exit (exit from the debugger)
    • Meta Commands
      • .help (show the help of commands)
      • .debug (prepare and connect to debugger)
      • .connect (connect to a session)
      • .disconnect (disconnect from a session)
      • .listen (listen on a port and wait for the debugger to connect)
      • .status (show the debugger status)
      • .start (start a new process)
      • .restart (restart the process)
      • .attach (attach to a process)
      • .detach (detach from the process)
      • .switch (show the list and switch between active debugging processes)
      • .kill (terminate the process)
      • .process, .process2 (show the current process and switch to another process)
      • .thread, .thread2 (show the current thread and switch to another thread)
      • .pagein (bring the page into the RAM)
      • .dump (save the virtual memory into a file)
      • .formats (show number formats)
      • .script (run batch script commands)
      • .sympath (set the symbol server)
      • .sym (load pdb symbols)
      • .pe (parse PE file)
      • .logopen (open log file)
      • .logclose (close log file)
      • .cls (clear the screen)
    • Extension Commands
      • !a (assemble physical address)
      • !pte (display page-level address and entries)
      • !db, !dc, !dd, !dq (read physical memory)
      • !eb, !ed, !eq (edit physical memory)
      • !sb, !sd, !sq (search physical memory)
      • !u, !u64, !u2, !u32 (disassemble physical address)
      • !dt (display and map physical memory to structures)
      • !track (track and map function calls and returns to the symbols)
      • !epthook (hidden hook with EPT - stealth breakpoints)
      • !epthook2 (hidden hook with EPT - detours)
      • !monitor (monitor read/write/execute to a range of memory)
      • !syscall, !syscall2 (hook system-calls)
      • !sysret, !sysret2 (hook SYSRET instruction execution)
      • !mode (detect kernel-to-user and user-to-kernel transitions)
      • !cpuid (hook CPUID instruction execution)
      • !msrread (hook RDMSR instruction execution)
      • !msrwrite (hook WRMSR instruction execution)
      • !tsc (hook RDTSC/RDTSCP instruction execution)
      • !pmc (hook RDPMC instruction execution)
      • !vmcall (hook hypercalls)
      • !exception (hook first 32 entries of IDT)
      • !interrupt (hook external device interrupts)
      • !dr (hook access to debug registers)
      • !ioin (hook IN instruction execution)
      • !ioout (hook OUT instruction execution)
      • !hide (enable transparent-mode)
      • !unhide (disable transparent-mode)
      • !measure (measuring and providing details for transparent-mode)
      • !va2pa (convert a virtual address to physical address)
      • !pa2va (convert physical address to virtual address)
      • !dump (save the physical memory into a file)
      • !pcitree (show PCI/PCIe device tree)
      • !pcicam (dump the PCI/PCIe configuration space)
      • !idt (show Interrupt Descriptor Table entries)
      • !apic (dump local APIC entries in XAPIC and X2APIC modes)
      • !ioapic (dump I/O APIC)
    • Scripting Language
      • Assumptions & Evaluations
      • Variables & Assignments
      • Casting & Type-awareness
      • Conditionals & Loops
      • Constants & Functions
      • Debugger Script (DS)
      • Examples
        • view system state (registers, memory, variables)
        • change system state (registers, memory, variables)
        • trace function calls
        • pause the debugger conditionally
        • conditional breakpoints and events
        • patch the normal sequence of execution
        • access to a shared variable from different cores
        • count occurrences of events
      • Functions
        • debugger
          • pause
        • events
          • event_enable
          • event_disable
          • event_clear
          • event_sc
          • event_inject
          • event_inject_error_code
          • flush
        • exports
          • print
          • printf
        • interlocked
          • interlocked_compare_exchange
          • interlocked_decrement
          • interlocked_exchange
          • interlocked_exchange_add
          • interlocked_increment
        • memory
          • check_address
          • eb, ed, eq
          • eb_pa, ed_pa, eq_pa
          • memcpy
          • memcpy_pa
          • memcmp
          • virtual_to_physical
          • physical_to_virtual
        • diassembler
          • disassemble_len
          • disassemble_len32
        • spinlocks
          • spinlock_lock
          • spinlock_lock_custom_wait
          • spinlock_unlock
        • strings
          • strlen
          • wcslen
          • strcmp
          • strncmp
          • wcscmp
          • wcsncmp
    • Commands Map
  • Tips & Tricks
    • Considerations
      • Basic concepts in Intel VT-x
      • VMX root-mode vs VMX non-root mode
      • The "unsafe" behavior
      • Script engine in VMX non-root mode
      • Difference between process and thread switching commands
      • Accessing Invalid Address
      • Transparent Mode
    • Nested-Virtualization Environments
      • Supported Virtual Machines
      • Run HyperDbg on VMware
      • Run HyperDbg on Hyper-V
      • Supporting VMware/Hyper-V
      • VMware backdoor I/O ports
    • Misc
      • Event forwarding
      • Event short-circuiting
      • Event calling stage
      • Instant events
      • Message overflow
      • Customize build
        • Increase Communication Buffer Size
        • Number of EPT Hooks in One Page
        • Change Script Engine Limitations
      • Enable and disable events in Debugger Mode
      • Switch to New Process Layout
  • Contribution
    • Style Guide
      • Coding style
      • Command style
      • Doxygen style
    • Logo & Artworks
  • Design
    • Features
      • VMM (Module)
        • Control over NMIs
        • VMX root-mode compatible message tracing
        • Design of !epthook
        • Design of !epthook2
        • Design of !monitor
        • Design of !syscall & !sysret
        • Design of !exception & !interrupt
    • Debugger Internals
      • Events
      • Conditions
      • Actions
      • Kernel Debugger
        • Design Perspective
        • Connection
  • Links
    • Twitter
    • Telegram
    • Discord
    • Matrix
    • Mastodon
    • YouTube
    • hwdbg (Chip Debugger)
    • Doxygen
    • Contribution
Powered by GitBook
On this page
  • What do we mean by "safe" and "unsafe"?
  • Things we should never do
  • How can we access the memory?
  • The address is valid, but HyperDbg can't read it!
Edit on GitHub
  1. Tips & Tricks
  2. Considerations

The "unsafe" behavior

What is "unsafe" behavior in using HyperDbg

PreviousVMX root-mode vs VMX non-root modeNextScript engine in VMX non-root mode

Last updated 3 months ago

What do we mean by "safe" and "unsafe"?

The term “safe” and “unsafe” is used a lot in HyperDbg. By “safe,” we mean something that works all the time and won’t cause a system crash or system halt.

It’s so tricky to manage code in the VMX root-mode. As HyperDbg gives you the ability to run your custom assembly code in all modes of execution, you should avoid doing some “unsafe” behavior that leads to system instability.

It is because, in the VMX root-mode, interrupts are masked (disabled), or transfer buffer from the VMX root-mode to the VMX non-root mode needs extra effort, and we should be cautious and avoid executing some APIs to be safe.

HyperDbg provides you a way of safely accessing a non-paged pool in user, kernel, and VMX root, and it sends the buffer to the user mode and HyperDbg itself in a safe manner.

Things we should never do

  • You should never use an NT API in the VMX-root if it has IRQL limitations. Make sure that you use only those APIs that are marked with (ANY LEVEL) in MSDN. For example, calling ExAllocatePoolTag will halt the system most of the time. Also, sometimes it's observed that even "ANY LEVEL" functions led to a system fault.

  • You should not access a paged pool while you're executing in the VMX root-mode. If you access these pages and the page is paged out, it'll result in a system halt or crash. It is because interrupts and exceptions are masked off in the VMX-root mode, and a page-fault exception will never deliver.

  • When you're executing in the VMX-root, your cr3 view of memory is the SYSTEM process (pid=4)'s view of memory. If you want to access another process memory, you should change the cr3 by yourself and return to system cr3 immediately. Otherwise, you won't let the HyperDbg return gracefully, and you'll see BSOD.

  • Do not directly access user mode codes from the VMX-root mode. It is because user-mode codes are located on paged memory, and there are two problems here. First, you need to find the cr3 of the user, and it's not the process cr3 because, for KPTI mitigation of Meltdown, cr3 of user space is different from kernel cr3. It's also observed that accessing a user-mode memory (even if it's recently executed) causes a system crash. The problem is solved if you inject a #PF to the guest with that address. Hence, it is maybe because Intel has different caches (instruction & data) for its internal use, and the page-fault for instruction and data is different, so executing an instruction doesn't mean that it's paged-in for data (read/write) access.

How can we access the memory?

  • We can access the memory safely by using the functions provided by HyperDbg commands and HyperDbg's script engine. For example, if we want to read memory from the current process, we can use the regular , and modify the memory using the commands. On the other hand, in the script engine, you can use different memory reading to read the memory or use the functions to modify the memory.

  • Other than that, you can use the function to make sure whether an address is valid and safe to access or not.

The address is valid, but HyperDbg can't read it!

  • Sometimes, addresses that you try to access are valid but HyperDbg is not able to read or modify the address and keeps showing a message that the address is invalid. This is mainly because the address is paged out and is not available in RAM. Or in some cases, the address is actually available in the RAM but the operating system didn't make it present for the current process's memory layout. The second case happens mostly when you start a new process.

  • In order to solve that, we could use the '' command to bring the pages into the RAM and make it available for both the process and HyperDbg.

db, dc, dd, dq (read virtual memory)
eb, ed, eq (edit virtual memory)
keywords
eb, ed, eq
check_address
.pagein