rocker9527 / windows-vs-linux-loader-architecture

Side-by-side comparison of the Windows and Linux (GNU) Loaders

Windows vs Linux Loader Architecture

This repo was released in conjunction with an article on "What is Loader Lock?".

The loader is a vital part of any operating system. It's responsible for loading programs and libraries into a process' address space and is the first component to run code when a process starts. Starting a process includes tasks like initializing critical data structures, loading dependencies, and running the program.

The intentions of this document are to:

1. Compare the Windows, Linux, and sometimes MacOS loaders
   • Provide perspective on architectural and ecosystem differences as well as how they coincide with the loader
   • Including experiments on how flexible or rigid they are with what can safely be done during module initialization (with the loader's internal locks held)
2. Formally document how a modern Windows loader performs synchronization
   • Current open source Windows implementations, including Wine and ReactOS, perform locking similar to the legacy Windows loader
3. Educate, satisfy curiosity, and help fellow reverse engineers
   • If you're looking for information on anything in particular, just give this document a CTRL+F

All of the information contained here covers Windows 10 22H2 and glibc 2.38. Some sections of this document also touch on the MacOS loader.

Table of Contents

• Windows vs Linux Loader Architecture
  • Table of Contents
  • Module Information Data Structures
  • Locks
  • Atomic state
  • State
  • Concurrency Experiments
  • Concurrency Experiments Conclusion
  • Comparing OS Library Loading Locations
  • DllMain vs Constructors and Destructors
  • Lazy Loading and Lazy Binding OS Comparison
  • LDR_DDAG_NODE.State Analysis
  • Symbol/Export Lookup Comparison (Windows GetProcAddress vs POSIX dlsym GNU Implementation)
    • How Does GetProcAddress/dlsym Handle Concurrent Library Unload?
  • GNU Loader Lock Hierarchy
  • Lazy Linking Synchronization
  • Loader Enclaves
    • The Windows LdrpObtainLockedEnclave Function
  • Modern Windows Loader Synchronization: State > Locks
    • Case Study
  • Windows Loader Initialization Locking Requirements
  • ELF Flat Symbol Namespace (GNU Namespaces and STB_GNU_UNIQUE)
  • What's the Difference Between Dynamic Linking, Static Linking, Dynamic Loading, Static Loading?
  • LoadLibrary vs dlopen Return Type
  • DllMain (Loader Lock) + COM and CLR
    • A Bonus Case Study on ABBA Deadlock Between the Loader Lock and GDI+ Lock
  • LdrpDrainWorkQueue and LdrpProcessWork
  • LdrpLoadCompleteEvent and LdrpWorkCompleteEvent
  • When the Process Would Rather Terminate Then Wait on a Critical Section
  • ABBA Deadlock
  • ABA Problem
  • Dining Philosophers Problem
  • Compiling & Running
  • Analysis Commands
    • LDR_DATA_TABLE_ENTRY Analysis
    • LDR_DDAG_NODE Analysis
    • Trace TEB.SameTebFlags
    • Searching Assembly for Structure Offsets
    • Monitor a Critical Section Lock
    • Track Loader Events
    • Disable Loader Worker Threads
    • List All LdrpCriticalLoaderFunctions
    • Get TCB and Set GSCOPE watchpoint (GNU Loader)
  • Debugging Notes
  • Big thanks to Microsoft for successfully nerd sniping me!

Module Information Data Structures

A Windows module list is a circular doubly linked list of type LDR_DATA_TABLE_ENTRY. The loader maintains multiple lists containing the same entries but in different link orders. These lists include InLoadOrderModuleList, InMemoryOrderModuleList, and InInitializationOrderModuleList. These list heads are stored in ntdll!PEB_LDR_DATA. Each LDR_DATA_TABLE_ENTRY structure houses a LIST_ENTRY structure (containing both Flink and Blink pointers) for each module list. See the "What is Loader Lock?" article for information on all the Windows loader data structures.

The Linux (glibc) module list is a non-circular doubly linked list of type link_map. link_map contains both the next and prev pointers used to link modules together into a list.

Locks

OS-level synchronization mechanisms of all kinds, including thread synchronization mechanisms (i.e. Windows critical section or POSIX mutex), exclusive/write and shared/read locks, and other synchronization objects (Windows). OS-level, meaning these locks may make a system call (a syscall instruction on x86-64) to perform a non-busy wait if the synchronization mechanism is owned/contended/waiting.

Some lock varieties are a mix of an OS lock and a spinlock (i.e. busy loop). For example, both a Windows critical section and a GNU mutex (not POSIX; this is a GNU extension) support specifying a spin count. A spin count is a potential performance optimization for avoiding expensive context switches between user mode and kernel mode.

Most OS locks use atomic flags to implement locking primitives when there's no contention (e.g. implemented with the lock cmpxchg instruction on x86). An exception to this would be Win32 event synchronization objects, which rely entirely on system calls (i.e. NtSetEvent and NtResetEvent functions are just stubs containing a syscall instruction).

Windows LdrpModuleDatatableLock

• Performs full blocking (exclusive/write) access to its respective module information data structures
  • This includes two linked lists (InLoadOrderModuleList and InMemoryOrderModuleList), a hash table, two red-black trees, and two directed acyclic graphs
    • Note: The DAGs only require LdrpModuleDatatableLock protection to ensure synchronization/consistency with other module information data structures during write operations (e.g. adding/deleting nodes)
      • In addition to acquiring the LdrpModuleDatatableLock lock, safely modifying the DAGs requires that your thread be the loader owner (LoadOwner in TEB.SameTebFlags and incrementing ntdll!LdrpWorkInProgress)
      • Read operations (e.g. walking the DAGs) are naturally protected depending on where the load owner is in the loading process (see: LDR_DDAG_NODE.State Analysis)
  • This lock also protects some structure members contained within these data structures (e.g. the LDR_DDAG_NODE.LoadCount reference count)
• Windows shortly acquires LdrpModuleDatatableLock many times (I counted 17 exactly) for every LoadLibrary (tested with a full path to an empty DLL; a completely fresh Visual Studio DLL project)
  • Acquiring this lock so many times could create contention on LdrpModuleDatatableLock, even if the lock is only held for short sprints
  • Monitor changes to LdrpModuleDatatableLock by setting a watchpoint: ba w8 ntdll!LdrpModuleDatatableLock (ensure you don't count unlocks)
    • Note: There are a few occurrences of this lock's data being modified directly for unlocking instead of calling RtlReleaseSRWLockExclusive (this is likely done as a performance optimization on hot paths)
• Implemented as a slim read/write (SRW) lock, typically acquired exclusively

Linux (GNU loader) _rtld_global._dl_load_write_lock

• Performs full blocking (exclusive/write) access to its respective module data structures
• On Linux, this is only a linked list (thelink_map list)
• Linux shortly acquires _dl_load_write_lock once on every dlopen from the _dl_add_to_namespace_list internal function (see GDB log for evidence of this)
  • Other functions that acquire _dl_load_write_lock (not called during dlopen) include the dl_iterate_phdr function which is for iterating over the module list
    • According to glibc source code, this lock is acquired to: "keep __dl_iterate_phdr from inspecting the list of loaded objects while an object is added to or removed from that list."
    • On Windows, acquiring the equivalent LdrpModuleDatatableLock is required to iterate the module list safely (e.g. when calling the LdrpFindLoadedDllByNameLockHeld function)

Windows LdrpLoaderLock

• Protects the InInitializationOrderModuleList linked list and blocks concurrent initialization/deinitialization modules
  • At process exit, RtlExitUserProcess acquires loader lock before running library deinitialization functions (DLL_PROCESS_DETACH).
  • The latter protection is also necessary because libraries that depend on each other cannot have their initialization functions run concurrently. Otherwise, a concurrent thread running a library's initialization function could call into an uninitialized dependency. Guarding against this is especially necessary on Windows, where the architecture prioritizes libraries that provide functionality over individual processes. Thus, loader lock also protects against concurrent out-of-order library initialization.
• Implemented as a critical section

Linux (GNU loader) _dl_load_lock

• This lock is acquired right at the start of a _dl_open, dlclose, and other loader functions
  • dlopen eventually calls _dlopen after some preparation work (which shows in the call stack) like setting up an exception handler
  • At this point, dlopen is committed to doing some loader work
• According to glibc source code, this lock's purpose is to: "Protect against concurrent loads and unloads."
  • This protection includes concurrent module initialization similar to how a modern Windows ntdll!LdrpLoaderLock does
  • For example, _dl_load_lock protects a concurrent dlclose from running a library's destructor before that library's constructor has finished running
  • _dl_load_lock (unlike Windows ntdll!LdrpLoaderLock) provides broad protection of the entire library loading/unloading process from start to end (including the only time a library is added/removed to/from the link_map), thus naturally allowing it to protect from modifications to the link_map module list as a result of the inherent lock hierarchy

Linux (GNU loader) _ns_unique_sym_table.lock

• This is a per-namespace lock for protecting that namespace's unique (GNU_STB_UNIQUE) symbol hash table
• GNU_STB_UNIQUE symbols are a type of symbol that a module can expose; they are considered a misfeature of the GNU loader
• As standardized by the ELF executable format, the GNU loader uses a per-module statically allocated (at compile time) symbol table for locating symbols within a module (.so shared object file); however, the _ns_unique_sym_table.lock lock protects a separate dynamically allocated hash table specially for GNU_STB_UNIQUE symbols
  • In Windows terminology, the closest approximation to a symbol would be a DLL's function exports (there's no mention of the word "export" in the objdump manual)
• You can use objdump -t to dump all the symbols, including unique symbols, of an ELF file
  • objdump -t displays unique global symbols (as the manual references them) with the u flag character
• Internally, the call chain for looking up a GNU_STB_UNIQUE symbol starting with dlsym goes dlsym -> dl_lookup_symbol_x -> do_lookup_x -> do_lookup_unique where finally, _ns_unique_sym_table.lock is acquired
• For more info on GNU_STB_UNIQUE symbols, see the GNU_STB_UNIQUE section

This is where the loader locks end for the GNU loader on Linux. Other than that, there's only a lock specific to thread-local storage (TLS) if your module uses that, as well as global scope (GSCOPE) locking.

However, Windows has more synchronization objects that control the loader, including:

• Loader events (Win32 event objects), including:
  • ntdll!LdrpInitCompleteEvent
    • This event being set indicates process initialization is complete
    • LdrpInitialize (_LdrpInitialize) creates this event
      • This event is created before process initialization
    • Thread startup waits on this
    • This event is only set (NtSetEvent) by the LdrpProcessInitializationComplete function soon after LdrpInitializeProcess returns, at which point it's never set/unset again
    • This event is created in the nonsignaled (i.e. waiting) state
  • ntdll!LdrpLoadCompleteEvent
    • Created by LdrpInitParallelLoadingSupport calling LdrpCreateLoaderEvents
    • Thread startup waits on this
    • This event is set in LdrpDropLastInProgressCount before relinquishing control as the LoadOwner
    • This event is created in the nonsignaled (i.e. waiting) state
  • LdrpWorkCompleteEvent
    • Created by LdrpInitParallelLoadingSupport calling LdrpCreateLoaderEvents
    • This event may be set (NtSetEvent) immediately before the LdrpProcessWork function returns
    • This event is created in the nonsignaled (i.e. waiting) state
  • All of these events are created by NTDLL at process startup using NtCreateEvent
  • For in-depth information on the latter two events, see the section on LdrpLoadCompleteEvent vs LdrpWorkCompleteEvent
• ntdll!LdrpWorkQueueLock
  • Implemented as a critical section
  • Used in LdrpDrainWorkQueue so only one thread can access the LdrpWorkQueue queue at a time
• ntdll!LdrpDllNotificationLock
  • This is a critical section; it's typically locked (RtlEnterCriticalSection) in the ntdll!LdrpSendPostSnapNotifications function (called by ntdll!PrepareModuleForExecution -> ntdll!LdrpNotifyLoadOfGraph; this happens in advance of ntdll!PrepareModuleForExecution calling LdrpInitializeGraphRecurse to do module initialization
  • The LdrpSendPostSnapNotifications function acquires this lock to ensure consistency with other functions that run DLL notification callbacks before fetching app compatibility data (SbUpdateSwitchContextBasedOnDll function) and potentially running post snap DLL notification callbacks (look for call qword ptr [ntdll!__guard_dispatch_icall_fptr (<ADDRESS>)] in disassembly)
    • Typically, LdrpSendPostSnapNotifications doesn't run any callback functions
    • Other functions that directly send DLL notifications: LdrpSendShimEngineInitialNotifications (called by LdrpLoadShimEngine and LdrpDynamicShimModule)
  • The ntdll!LdrpSendNotifications function (called by LdrpSendPostSnapNotifications) function recursively acquires this lock to safely access the LdrpDllNotificationList list so it can call the callback functions stored inside
    • By default, the LdrpDllNotificationList list is empty (so the LdrpSendDllNotifications function doesn't send any callbacks)
    • Notifications callbacks are registered with LdrRegisterDllNotification and are then sent with LdrpSendDllNotifications (it runs the callback function)
    • Functions calling DLL notification callbacks must hold the LdrpDllNotificationLock lock during callback execution. This is similar to how executing module initialization/deinitialization code (DllMain) requires holding the LdrpLoaderLock lock.
    • Other functions that may call LdrpSendDllNotifications include: LdrpUnloadNode and LdrpCorProcessImports (may be called by LdrpMapDllWithSectionHandle)
    • LdrpSendDllNotifications takes a pointer to a LDR_DATA_TABLE_ENTRY structure as its first argument
    • In ReactOS, LdrpSendDllNotifications is referenced in LdrUnloadDll sending a shutdown notification with a FIXME comment (not implemented yet): //LdrpSendDllNotifications(CurrentEntry, 2, LdrpShutdownInProgress);
      • ReactOS code analysis: If LdrpShutdownInProgress is set, LdrUnloadDll skips deinitialization and releases loader lock early, presumably so shutdown happens faster (don't need to deinitialize a module if the whole process is about to not exist)
      • LdrpShutdownInProgress is referenced in PEB_LDR_DATA
  • Reading loader disassembly, you may see quite a few places where loader functions check the LdrpDllNotificationLock lock like so: RtlIsCriticalSectionLockedByThread(&LdrpDllNotificationLock)
    • For instance, in the ntdll!LdrpAllocateModuleEntry, ntdll!LdrGetProcedureAddressForCaller, ntdll!LdrpPrepareModuleForExecution, and ntdll!LdrpMapDllWithSectionHandle functions
    • These checks detect if the current thread is executing a DLL notification callback and then implement special logic for that edge case (for this reason, they can generally be ignored)
    • The exact differing logic varies widely depending on the function checking
    • By putting Google Chrome under WinDbg, I found an instance where the loader ran a post-snap DLL notification callback. The callback ran the apphelp!SE_DllLoaded function.
    • Lastly, ignore any AVrf* prefixed functions as this only ties in with debugging functionality provided by Application Verifier
• PEB.TppWorkerpListLock
  • This SRW lock (typically acquired exclusively) exists in the PEB to control access to the member immediately below it, which is the TppWorkerpList doubly linked list
  • This list keeps track of all the threads belonging to any thread pool in the process
    • These threads show up as ntdll!TppWorkerThread threads in WinDbg
    • There's a list head, after which each LIST_ENTRY points into the stack memory a thread within the ntdll!TppWorkerThread function's stack frame
      • The TpInitializePackage function (called by LdrpInitializeProcess) initializes the list head, then the main ntdll!TppWorkerThread function of each new thread belonging to a thread pool adds itself to the list
    • The threads in this list include threads belonging to the loader worker thread pool (LoaderWorker in TEB.SameTebFlags)
• LDR_DATA_TABLE_ENTRY.Lock
  • Starting with Windows 10, each LDR_DATA_TABLE_ENTRY has a PVOID Lock member (it replaced a Spare slot)
  • LdrpWriteBackProtectedDelayLoad uses this per-node lock to implement some level of protection during delay loading
    • This function calls NtProtectVirtualMemory, so it's likely something to do with the setting memory protection on a module's Import Address Table (IAT)
• Searching symbols reveals more locks: x ntdll!Ldr*Lock
  • LdrpDllDirectoryLock, LdrpTlsLock (this is an SRW lock typically acquired in the shared mode), LdrpEnclaveListLock, LdrpPathLock, LdrpInvertedFunctionTableSRWLock, LdrpVehLock, LdrpForkActiveLock, LdrpCODScenarioLock, LdrpMrdataLock, and LdrpVchLock
  • A lot of these locks seem to be for controlling access to list data structures

Atomic state

An atomic state value is modified using a single assembly instruction. On an SMP operating system (e.g. Windows and very likely your build of Linux; check with uname -a command) with a multi-core processor, this must include a lock prefix (on x86) so the processor knows to synchronize that memory access across CPU cores. The x86 ISA requires that a single read/write operation is atomic by default. It's only when atomically combining multiple reads or writes (e.g. to atomically increment a reference count) that the lock prefix is necessary. Only a few key pieces of the Windows loader atomic state I came across are listed here.

• ntdll!LdrpProcessInitialized
  • This value is modified atomically with a lock cmpxchg instruction
  • As the name implies, it indicates whether process initialization has been completed (LdrpInitializeProcess)
  • This is an enum ranging from zero to three; here are the state transitions:
    • NTDLL compile time initialization starts LdrpProcessInitialized with a value of zero (process is uninitialized)
    • LdrpInitialize increments LdrpProcessInitialized to one zero early on (initialization event created)
      • If the process is still initializing, newly spawned threads jump to calling NtWaitForSingleObject, waiting on the LdrpInitCompleteEvent loader event before proceeding
      • Before the loader calls NtCreateEvent to create LdrpInitCompleteEvent at process startup, spawning a thread into a process causes it to use LdrpProcessInitialized as a spinlock (i.e. a busy loop)
      • For example, if a remote process calls CreateRemoteThread and the thread spawns before the creation of LdrpInitCompleteEvent (an unlikely but possible race condition)
    • LdrpProcessInitializationComplete increments LdrpProcessInitialized to two (process initialization is done)
      • This happens immediately before setting the LdrpInitCompleteEvent loader event so other threads can run
      • After LdrpProcessInitializationComplete returns, NtTestAlert processes the asynchronous procedure call (APC) queue, and finally, NtContinue yields code execution of the current thread to KERNEL32!BaseThreadInitThunk, which eventually runs our program's main function
• LDR_DATA_TABLE_ENTRY.ReferenceCount
  • This is a reference count for LDR_DATA_TABLE_ENTRY structures
  • On load/unload, this state is modified by lock inc/dec instructions, respectively (although, on the initial allocation of a LDR_DATA_TABLE_ENTRY before linking it into any shared data structures, of course, no locking is necessary)
  • The LdrpDereferenceModule function atomically decrements LDR_DATA_TABLE_ENTRY.ReferenceCount by passing 0xffffffff to a lock xadd assembly instruction, causing the 32-bit integer to overflow to one less than what it was (x86 assembly doesn't have an xsub instruction so this is the standard way of doing this)
    • Note that the xadd instruction isn't the same as the add instruction because the former also atomically exchanges (hence the "x") the previous memory value into the source operand
      • This is at the assembly level; in code, Microsoft is likely using the InterlockedExchangeSubtract macro to do this
    • The LdrpDereferenceModule function tests (among other things) if the previous memory value was 1 (meaning post-decrement, it's now zero; i.e. nobody is referencing this LDR_DATA_TABLE_ENTRY anymore) and takes that as a cue to unmap the entire module from memory (calling the LdrpUnmapModule function, deallocating memory structures, etc.)
• ntdll!LdrpLoaderLockAcquisitionCount
  • This value is modified atomically with the lock xadd prefixed instructions
  • It was only ever used as part of cookie generation in the LdrLockLoaderLock function
    • On both older/modern loaders, LdrLockLoaderLock adds to LdrpLoaderLockAcquisitionCount every time it acquires the loader lock (it's never decremented)
    • In a legacy (Windows Server 2003) Windows loader, the LdrLockLoaderLock function (an NTDLL export) was often used internally by NTDLL even in Ldr prefixed functions. However, in a modern Windows loader, it's mostly phased out in favor of the LdrpAcquireLoaderLock function
    • In a modern loader, the only places where I see LdrLockLoaderLock called are from non-Ldr prefixed functions, specifically: TppWorkCallbackPrologRelease and TppIopExecuteCallback (thread pool internals, still in NTDLL)

State

A state value has access to it protected by one of the aforementioned locks or doesn't require protection. Only a few key pieces of Windows loader state I came across are listed here.

• LDR_DDAG_NODE.State (pointed to by LDR_DATA_TABLE_ENTRY.DdagNode)
  • Each module has a LDR_DDAG_NODE structure with a State member containing 15 possible states -5 through 9
  • LDR_DDAG_NODE.State tracks a module's entire lifetime from allocating module information data structures (LdrpAllocatePlaceHolder) and loading to unload and subsequent deallocation of its module information structures
    • In my opinion, this makes the combined LDR_DDAG_NODE.State values of all modules the most important piece of loader state
  • It requires parallel loader initialization before use (LdrpInitParallelLoadingSupport), so it plays no part in ntdll.dll loading
  • Performing each state change may necessitate acquiring the LdrpModuleDatatableLock lock to ensure consistency between module information data structures
    • The specific state changes requiring LdrpModuleDatatableLock protection are documented in the link below
  • Please see the LDR_DDAG_NODE.State Analysis for more info
• ntdll!LdrpWorkInProgress
  • This reference count is a key piece of loader state (zero meaning work is not in progress and up from that meaning work is in progress)
    • It's not modified atomically with lock prefixed instructions
    • "Work" here refers to load owner work
  • Acquiring the LdrpWorkQueueLock lock is a requirement for safely modifying the LdrpWorkInProgress state
    • I verified this by setting a watchpoint on LdrpWorkInProgress and noticing that LdrpWorkQueueLock is always locked while checking/modifying the LdrpWorkInProgress state (also, I searched disassembly code)
      • Acquiring this lock ensures consistency throughout the loader work functions (e.g. LdrpDrainWorkQueue and LdrpProcessWork functions) and ntdll!LdrpWorkQueue linked list before modifying the shared state
      • The LdrpDropLastInProgressCount function makes this clear because it briefly acquires LdrpWorkQueueLock only around the single assembly instruction that sets LdrpWorkInProgress to zero
• ntdll!LdrInitState
  • This value is not modified atomically with lock prefixed instructions
    • Loader initialization is a procedural process only occurring once and on one thread, so this value doesn't require protection
  • In ReactOS code, the equivalent value is LdrpInLdrInit, which the code declares as a BOOLEAN value
  • In a modern Windows loader, this a 32-bit integer (likely an enum) ranging from zero to three; here are the state transitions:
    • LdrpInitialize initializes LdrInitState to zero (loader is uninitialized)
    • LdrpInitializeProcess calls LdrpEnableParallelLoading and immediately after sets LdrInitState to one (import loading in progress; thank you, Windows Internals book authors)
    • LdrpInitializeProcess calls LdrpPrepareModuleForExecution on the EXE (first argument is LDR_DATA_TABLE_ENTRY of EXE) and immediately after sets LdrInitState to two (imports have loaded)
      • Note: LdrpPrepareModuleForExecution does not call LdrpInitializeGraphRecurse in this scenario because our EXE isn't a DLL with a DllMain initialization function but still calls other functions like LdrpNotifyLoadOfGraph and LdrpDynamicShimModule
    • LdrpInitialize (LdrpInitializeProcess returned), shortly before calling LdrpProcessInitializationComplete, sets LdrInitState to three (loader initialization is done)
• LDR_DDAG_NODE.LoadCount
  • This is the reference count for a LDR_DDAG_NODE structure; safely modifying it requires acquiring the LdrpModuleDataTableLock lock
• TEB.WaitingOnLoaderLock is thread-specific data set when a thread is waiting for loader lock
  • RtlpWaitOnCriticalSection (RtlEnterCriticalSection calls RtlpEnterCriticalSectionContended, which calls this function) checks if the contended critical section is LdrpLoaderLock and if so, sets TEB.WaitingOnLoaderLock equal to one
    • This branch condition runs every time any contended critical section gets waited on, which is interesting (monolithic much?)
  • RtlpNotOwnerCriticalSection (called from RtlLeaveCriticalSection) also checks LdrpLoaderLock (and some other info from PEB_LDR_DATA) for special handling
    • However, this is only for error handling and debugging because a thread that doesn't own a critical section should have never attempted to leave it in the first place
• Flags in TEB.SameTebFlags, including: LoadOwner, LoaderWorker, and SkipLoaderInit
  • All of these were introduced in Windows 10 (SkipLoaderInit only in 1703 and later)
  • LoadOwner (flag mask 0x1000) is state that a thread uses to inform itself that it's the one responsible for completing the work in progress (ntdll!LdrpWorkInProgress)
    • The LdrpDrainWorkQueue function sets the LoadOwner flag on the current thread immediately after setting ntdll!LdrpWorkInProgress to 1, thus directly connecting these two pieces of state
    • The LdrpDropLastInProgressCount function unsets this flag along with ntdll!LdrpWorkInProgress
    • Any thread doing loader work (e.g. LoadLibrary) will temporarily receive this TEB flag
    • This state is local to the thread (in the TEB), so it doesn't require the protection of a lock
  • LoaderWorker (flag mask 0x2000) identifies loader worker threads
    • These show up as ntdll!TppWorkerThread in WinDbg
    • On thread creation, LdrpInitialize checks if the thread is a loader worker and, if so, handles it specially (for more info, see the "What is Loader Lock?" article)
    • Flag can be set on a new thread using NtCreateThreadEx
  • SkipLoaderInit (flag mask 0x4000) tells the spawning thread to skip all loader initialization
    • As per the Windows Internals book and confirmed by myself
      • In LdrpInitialize IDA decompilation, you can see SameTebFlags being tested for 0x4000, and if present, loader initialization is completely skipped (_LdrpInitialize is never called)
    • This could be useful for creating new threads without being blocked by loader events
    • Threads with the SkipLoaderInit flag can be created using NtCreateThreadEx

Indeed, the Windows loader is an intricate and monolithic state machine. Its complexity stands out when put side-by-side with the simple glibc loader on Linux. This complexity helps explain why process creation on Linux is faster than on Windows.

Concurrency Experiments

• Load libraries
  • We set hardware (write) watchpoints on the _dl_load_lock and _dl_load_write_lock data, then load two libraries to check how often and where in the code these locks are acquired
  • See the GDB log
• Loading a library from a library constructor (recursive loading)
  • Windows: ✔️
  • Linux: ✔️
• Lazy/delay loading of a library import from a library constructor
  • Windows: ✘ (deadlock)
    • Lazy loading may fail (even within the same thread), but lazy linking works
    • This hurdle came up in "Perfect DLL Hijacking"; look for the default delay load helper function: __delayLoadHelper2 in the deadlocked call stack
    • Microsoft documentation: "A DLL project that delays the loading of one or more DLLs itself shouldn't call a delay-loaded entry point in DllMain."
  • Linux: ✔️
    • Both dynamic loading (dlopen) and dynamic linking (at load time) tests were done in the most equivalent way
  • As mentioned in the Lazy Loading and Lazy Binding OS Comparison section, Linux doesn't natively support lazy loading
    • However, effectively accomplishing lazy loading through various techniques comes down to recursive loading, which is already known to work
    • What I tested in this experiment is that lazy binding works under _dl_load_lock
    • Regardless, the GNU loader is immune to lazy loading deadlocks, so it gets a pass
• Spawning a thread and then waiting for its creation/termination from a library constructor
  • Windows: ✘ (deadlock)
    • CreateThread then WaitForSingleObject on thread handle from DllMain
    • Deadlock on LdrpInitCompleteEvent in LdrpInitialize function during process startup (see Windows Loader Initialization Locking Requirements for more info) or deadlock on LdrpLoadCompleteEvent in LdrpInitializeThread (runs DLL_THREAD_ATTACH; this is essentially useless by the way and libraries often disable it, among other things) -> LdrpDrainWorkQueue function during process run-time
    • This represents an overextension of the loader's locks because the threading implementation should have zero knowledge of the loader (or at least its locking) and vice-versa
    • Specifying THREAD_CREATE_FLAGS_SKIP_LOADER_INIT (new in Windows 10) on NtCreateThreadEx can bypass these thread creation blockers
      • I haven't seen an occurrence of Windows internally spawning a thread with this flag, so these deadlocks remain prevalent
  • Linux: ✔️
• Registering an atexit handler from a library
  • Windows: ✘ (either loader lock or CRT exit lock)
    • In a DLL, VC++ compiles atexit in a DLL down running in the DLL_PROCESS_DETACH of that module, so there's loader lock
    • In an EXE, an atexit handler runs under the CRT "exit" critical section lock, which isn't thread safe
  • Linux: ✔️
    • At process exit, there's no loader lock (_dl_load_lock); if a programmer closes the library (dlclose) then there's loader lock
      • The POSIX atexit specification doesn't mention libraries and what should happen if the atexit handler gets unloaded from memory, but this seems like the best possible handling of the scenario
    • Releases the more specialized __exit_funcs_lock lock before calling into an atexit handler, then reacquires when the lock after
    • The __run_exit_handlers function was designed to be reentrant and thread-safe
• Spawning a thread from a library constructor, waiting on the thread's termination, then loading another library in the new thread
  • Windows: ✘ (deadlock, failed the prerequisite test)
  • Linux: ✘ (deadlock)
    • See the log, we deadlock while trying to acquire _dl_load_lock on the new thread
  • This failing makes sense because the thread loading lib1 already holds the loader lock. Consequently, waiting on the termination of a new thread that tries to acquire the loader lock so it can load lib2 will deadlock. Recursive acquisition of the loader lock can't happen because lib2 is being loaded on a separate thread. We created a deadlock in the loader.

Concurrency Experiments Conclusion

The Windows loader, in contrast to the GNU loader (or Unix-like loaders in general), is more vulnerable to deadlock or crash scenarios for a variety of architectural reasons, including:

• Windows is a monolith
  • The monolithic architecture of the Windows API may cause separate lock hierarchies (e.g. Windows loader and COM) to interleave between threads potentially causing a deadlock
  • The Windows threading implementation meshes with the loader at thread startup and exit
  • Windows kernel mode and user mode closely integrate (NT and NTDLL), whereas, on GNU/Linux, these two vital OS components aren't even made by the same people (glibc userland and Linux kernel all held together by, at minimum, the POSIX, System V ABI, and C standards)
  • Contrast this with the Unix philosophy, which encourages modularity
• Windows architecture prioritizes libraries that provide functionality, processes housing multiple threads, and remote procedure calls (RPC) over separate processes
  • Splitting core Windows API, C runtime, shell (SHELL32 and SHCORE libraries; accessing "the" shell through libraries isn't how it works on Unix-like systems), and lots of other functionality into many DLLs leaves more opportunities for actions done during library loading/unloading or initialization/deinitialization to go wrong
    • A function in the DllMain of a module could load a library (perhaps delay loading), thus calling into another module's DllMain while the former module is potentially partially initialized
    • A module you depend on or call a function from may have already been deinitialized by the time your module's DLL_PROCESS_DETACH is run
    • Microsoft introducing the dependency graph in newer Windows versions helped solve this problem for DLL_PROCESS_ATTACH, but there could potentially still be issues due to circular dependencies ("dependency loops") or GetModuleHandle then GetProcAddress (not dynamic linking) from DllMain assuming a DLL is already loaded (Nor POSIX or GNU loader extensions define a function like GetModuleHandle, instead only having dlopen due to the former function's inherent weakness)
    • Poses a greater risk of an application performing an action that assumes DLL_PROCESS_ATTACH and DLL_PROCESS_DETACH always run on the same thread when they may not (this is true for the GNU loader too, but Windows has a higher risk due to its architecture)
  • In contrast, Unix-like systems prioritize processes and the idea that everything is a file
    • All the core system functionality is exposed by the single libc implementation library, possibly with a separate ld library for the loader (although musl has these combined into a single libc library and glibc has recently done the same to improve process startup performance)
• Non-reentrant or thread-unsafe design
  • Notable Windows components like CRT/DLL exit (atexit) and FLS callbacks either weren't designed with reentrancy or thread safety in mind and may also run under loader lock (note: these aren't part of the loader but they may integrate with it so the point stands)
  • Delay loading is non-reentrant in some cases, potentially leading to deadlock
• Unexpectedly loading libraries
  • Inherently, delay loading may unexpectedly cause library loading when the programmer didn't intend, thus leading to deadlock
• Backward compatibility
  • Microsoft's intense commitment to staying backward compatible, often forever, means you have to remain bug-compatible with all your mistakes once they (whether you want them to or not) become part of a public API/ABI
  • With concurrency, this means you must be incredibly careful not to break any assumptions applications may make about thread synchronization, the order of operations, and related state
    • Also, if you want to completely redesign a system or data structure to unlock certain operations/functions and increase overall concurrency (perhaps switching to a careful design utilizing more atomic loads and stores), well, there's a good chance you can't due to backward compatibility
    • This fact rings especially true when DLLs like ntdll.dll put a pointer to its locks, such as loader lock, in the publicly accessible PEB or msvcrt.dll makes _lock and _unlock a part of their exported API; even if there is a leading underscore marking it as an implementation detail (the UCRT ucrtbase.dll DLL doesn't expose this functionality)
    • This fact also doesn't bode well with the Redmond-based business' reputation of "moving fast and breaking things" to achieve market dominance in a new space (although this strategy with Windows has subsided now, the technical debt remains)

The outcomes of these architectural facts and faults are far-reaching. Additionally, they often tie into each other, thus complicating matters.

Comparing OS Library Loading Locations

The Windows loader is searching for DLLs to load in a vast (and growing) number of places. Strangely, Windows uses the PATH environment variable for locating programs (similar to Unix-like systems) as well as DLLs. This Microsoft documentation still doesn't cover all the possible locations, though, because while debugging the loader during a LoadLibrary, I saw LdrpSendPostSnapNotifications eventually calls through to SbpRetrieveCompatibilityManifest (this isn't part of a notification callback). This Sbp-prefixed function searches for application compatibility shims in SDB files which may result in a compat DLL loading. Then there's also WinSxS and activation contexts. The plethora of possible search locations may contribute to DLL Hell and DLL hijacking (also known as DLL preloading or DLL sideloading) problems in Windows, the latter of which makes vulnerabilities due to a privileged process loading an attacker-controlled library more likely.

The GNU/Linux ecosystem differs due to system package managers (e.g. apt or dnf). All programs are built against the same system libraries (this is possible because all the packages are open source). Proprietary apps are generally statically linked or come with all their necessary libraries. The trusted directories for loading libraries can be found in the ldconfig manual. Beyond that, you can set the LD_LIBRARY_PATH environment variable to choose other places the loader should search for libraries, and binaries can include an rpath to specify additional run-time library search paths.

DllMain vs Constructors and Destructors

A constructor or destructor is the cross-platform equivalent to the Windows DllMain (called at DLL_PROCESS_ATTACH and DLL_PROCESS_DETACH time).

The Windows loader may call each module's DllMain at DLL_THREAD_ATTACH and DLL_THREAD_DETACH times. This only happens at thread start/exit time, meaning, for example, that a DLL loaded after thread start won't preempt that thread to run its DllMain with DLL_THREAD_ATTACH. These calls can be disabled as a performance optimization by calling DisableThreadLibraryCalls at DLL_PROCESS_ATTACH time.

Constructors and destructors originate from object-oriented programming standardized in C++ (C++ got initial standardization in 1998). They don't exist in the C standard. Constructors/destructors are what module initialization/deinitialization functions are known as in OOP languages. However, the concept of code that runs when a module loads/unloads goes at least as far back as System V Application Binary Interface Version 4. Compilers commonly provide access to the initialization/deinitialization functions with compiler-specific syntax. In GCC, these would be the __attribute__((constructor)) and __attribute__((destructor)) functions or the _init and _fini functions historically.

In Unix executable formats (e.g. Linux ELF and MacOS Mach-O), constructors and destructors are standardized in the System V Application Binary Interface as the .init and .fini sections. Note that this doesn't cover the non-standard but common and agreed-upon .init_array/.fini_array sections or the deprecated .ctors/.dtors sections, which modern binaries don't include (verified with objdump -h).

The PE (Windows) executable format standard doesn't define the .init and .fini sections; instead, a DllMain or any constructors/destructors are included with the rest of the program code in the .text section.

The Windows loader calls a module's LDR_DATA_TABLE_ENTRY.EntryPoint at module initialization; it has no knowledge of DllMain or constructors/destructors. Merging these into one callable EntryPoint is the job of a compiler (e.g. MSVC), which compiles a stub into your DLL that calls any constructors/destructors followed by DllMain.

Common Object File Format (COFF) is a common standard used as the basis for Portable Executable (PE), Executable and Linkable Format (ELF), and Mach object (Mach-O) executable formats. However, COFF only defines the text, data, and bss sections. COFF inherited these sections from its predecessor, a.out. COFF doesn't standardize section names (with the leading .); they are only a convention. Computers look for the STYP_TEXT, STYP_DATA, and STYP_BSS section header flags to identify a section (flag names may vary across executable formats). Section names are only there for human convenience and, especially on Windows, can be set arbitrarily. While COFF and PE don't mandate section names, compliance with the System V Application Binary Interface specification does reserve and require some sections to have specific names.

Windows has a .NET-supported (CLR) .cctor PE section.

Lazy Loading and Lazy Binding OS Comparison

It's important to differentiate between loading and binding. Loading involves mapping into memory. Binding refers to resolving symbols. Doing either of these operations lazily means that it's done on an as-needed basis instead of all at process startup or all at library load for the latter operation.

Windows collectively refers to lazy loading and lazy binding as "delay loading". However, I use the distinguished terms throughout this section.

Lazy loading exists as a first-class citizen on Windows but not on Unix systems. While it's a potentially useful optimization, architectural differences make it less needed on Unix systems than Windows. Windows prefers fewer programs with more threads, which get their functionality from libraries. In contrast, Unix favors more programs with fewer threads, leading to, on average, fewer libraries per process. Looking at the lengthy dependency chains for many common DLLs ties this in with the more monolithic architecture of Windows.

One can quickly check the dependency chain of a DLL by using dumpbin, which is a tool that ships with Visual Studio: dumpbin.exe /imports C:\Windows\System32\shell32.dll

Using dumpbin on kernel32.dll reveals it contains a lazy load for rpcrt4.dll when any of these also lazy binding imports are called:

Section contains the following delay load imports:

...

    RPCRT4.dll
              00000001 Characteristics
              6B8B06D0 Address of HMODULE
              6B8C0000 Import Address Table
              6B892B2C Import Name Table
              6B892C70 Bound Import Name Table
              00000000 Unload Import Name Table
                     0 time date stamp

          6B824205             167 RpcAsyncCompleteCall
          6B8241F1             20C RpcStringBindingComposeW
          6B8241E7             171 RpcBindingFromStringBindingW
          6B8241CC             169 RpcAsyncInitializeHandle
          6B8241FB              30 I_RpcExceptionFilter
          6B82420F             181 RpcBindingSetAuthInfoExW
          6B824237              99 NdrAsyncClientCall
          6B824223             166 RpcAsyncCancelCall
          6B82422D             16F RpcBindingFree
          6B824219             210 RpcStringFreeW

On Windows, lazy loading/binding is both a feature of the MSVC linker and the Windows loader with the NTDLL exported functions LdrResolveDelayLoadsFromDll, LdrResolveDelayLoadedAPI, and LdrQueryOptionalDelayLoadedAPI (more functions internally, not exposed as exports).

On Unix systems, lazy loading can be effectively achieved by doing dlopen and then dlsym at run-time. There are techniques to get more seamless lazy loading on Unix systems, but only with caveats. One could also employ a proxy design pattern in their code to achieve the effect. Mac used to support lazy loading until Apple removed this feature from the linker used by Xcode. Apple likely removed lazy loading because it's inherently vulnerable to deadlocking on loader lock. This vulnerability is due to lazy loading's ability to cause an application to load a library unexpectedly when calling a lazy-loaded function.

On the other hand, lazy linking (as it's referred to in the GNU ld manual) is a first-class citizen on both POSIX and Windows systems. On POSIX systems, this is simply calling dlopen with the RTLD_LAZY flag or providing the -z lazy flag to the linker. Windows doesn't seem to expose this functionality as readily (e.g. through LoadLibrary), but if nothing else, you always call the aforementioned NTDLL exports directly.

LDR_DDAG_NODE.State Analysis

LDR_DDAG_NODE.State or LDR_DDAG_STATE tracks a module's entire lifetime from beginning to end. With this analysis, I intend to extrapolate information based on the known types given to us by Microsoft (dt _LDR_DDAG_STATE command in WinDbg).

Findings were gathered by tracing all LDR_DDAG_STATE.State values at load-time and tracing a library unload, as well as searching disassembly.

Each state represents a stage of load owner work. This table comprehensively documents where these state changes occur throughout the loader and which locks are present. Doing load owner work requires first incrementing the ntdll!LdrpWorkInProgress reference counter. ntdll!LdrpWorkInProgress can only safely be decremented when the loader has completed all stages of loader work on the module(s) it's operating on. Only one thread, the load owner, can do loader work at a time. For instance, a typical load ranging from LdrModulesPlaceHolder to LdrModulesReadyToRun (may also include LdrModulesMerged) or a typical unload ranging from LdrModulesUnloading to LdrModulesUnloaded.

Note: Be sure to distinguish between load owner work (ntdll!LdrpWorkInProgress) with processing (mapping and snapping) work (involving ntdll!LdrpWorkCompleteEvent and ntdll!LdrpWorkQueue); this work can be parallelized with the LoaderWorker parallel loader threads. The naming can be a bit confusing due to the broad usage of "work". Any thread can enqueue work to the work queue for (potentially parallel; depending on MaxLoaderThreads) processing, then the next thread to become the load owner can do the rest of the work (not reading from ntdll!LdrpWorkQueue but other module information data structures like the DAG(s) for module initialization/deinitialization). The LdrpDrainWorkQueue function (only to be called by a thread who intends to become the load owner) ties the entire loading process together (see later the almost full reverse engineering of the LdrpDrainWorkQueue function).

LDR_DDAG_STATE States	State Changing Function(s)	Remarks
LdrModulesMerged (-5)	LdrpMergeNodes	LdrpModuleDatatableLock is held during this state change. See LdrModulesCondensed state for more info.
LdrModulesInitError (-4)	LdrpInitializeGraphRecurse	During DLL_PROCESS_ATTACH, if a module's DllMain returns FALSE for failure then this module state is set (any other return value counts as success). LdrpLoaderLock is held here.
LdrModulesSnapError (-3)	LdrpCondenseGraphRecurse	This function may set this state on a module if a snap error occurs. See LdrModulesCondensed state for more info.
LdrModulesUnloaded (-2)	LdrpUnloadNode	Before setting this state, LdrpUnloadNode may walk LDR_DDAG_NODE.Dependencies, holding LdrpModuleDataTableLock to call LdrpDecrementNodeLoadCountLockHeld thus decrementing the LDR_DDAG_NODE.LoadCount of dependencies and recursively calling LdrpUnloadNode to unload dependencies. Loader lock (LdrpLoaderLock) is held here.
LdrModulesUnloading (-1)	LdrpUnloadNode	Set near the start of this function. This function checks for LdrModulesInitError, LdrModulesReadyToInit, and LdrModulesReadyToRun states before setting this new state. After setting state, this function calls LdrpProcessDetachNode. Loader lock (LdrpLoaderLock) is held here.
LdrModulesPlaceHolder (0)	LdrpAllocateModuleEntry	The loader directly calls LdrpAllocateModuleEntry until parallel loader initialization (LdrpInitParallelLoadingSupport) occurs at process startup. At which point (with exception to directly calling LdrpAllocateModuleEntry once more soon after parallel loader initialization to allocate a module entry for the EXE), the loader calls LdrpAllocatePlaceHolder (this function first allocates a LDRP_LOAD_CONTEXT structure), which calls through to LdrpAllocateModuleEntry (this function places a pointer to this module entry's LDRP_LOAD_CONTEXT at LDR_DATA_TABLE_ENTRY.LoadContext). LdrpAllocateModuleEntry, along other like creating the module's LDR_DATA_TABLE_ENTRY, allocates a LDR_DDAG_NODE with zero-initialized heap memory.
LdrModulesMapping (1)	LdrpMapCleanModuleView	I've never seen this function get called; state typically jumps from 0 to 2. Only the LdrpGetImportDescriptorForSnap function may call this function which itself may only be called by LdrpMapAndSnapDependency (according to a disassembly search). LdrpMapAndSnapDependency typically calls LdrpGetImportDescriptorForSnap; however, LdrpGetImportDescriptorForSnap doesn't typically call LdrpMapCleanModuleView. This state is set before mapping a memory section (NtMapViewOfSection). Mapping is the process of loading a file from disk into memory.
LdrModulesMapped (2)	LdrpProcessMappedModule	LdrpModuleDatatableLock is held during this state change.
LdrModulesWaitingForDependencies (3)	LdrpLoadDependentModule	This state isn't typically set but, during a trace, I was able to observe the loader set it by launching a web browser (Google Chrome) under WinDbg, which triggered the watchpoint in this function when loading app compatibility DLL C:\Windows\System32\ACLayers.dll. Interstingly, the LDR_DDAG_STATE decreases by one here from LdrModulesSnapping to LdrModulesWaitingForDependencies; the only time I've observed this. LdrpModuleDatatableLock is held during this state change.
LdrModulesSnapping (4)	LdrpSignalModuleMapped or LdrpMapAndSnapDependency	In the LdrpMapAndSnapDependency case, a jump from LdrModulesMapped to LdrModulesSnapping may happen. LdrpModuleDatatableLock is held during state change in LdrpSignalModuleMapped, but not in LdrpMapAndSnapDependency. Snapping is the process of resolving the library’s import address table.
LdrModulesSnapped (5)	LdrpSnapModule or LdrpMapAndSnapDependency	In the LdrpMapAndSnapDependency case, a jump from LdrModulesMapped to LdrModulesSnapped may happen. LdrpModuleDatatableLock isn't held here in either case.
LdrModulesCondensed (6)	LdrpCondenseGraphRecurse	This function recevies a LDR_DDAG_NODE as its first argument and recursively calls itself to walk LDR_DDAG_NODE.Dependencies. On every recursion, this function checks whether it can remove the passed LDR_DDAG_NODE from the graph. If so, this function acquires LdrpModuleDataTableLock to call the LdrpMergeNodes function, which receives the same first argument, then releasing LdrpModuleDataTableLock after it returns. LdrpMergeNodes discards the uneeded node from the LDR_DDAG_NODE.Dependencies and LDR_DDAG_NODE.IncomingDependencies DAG adjacency lists of any modules starting from the given parent node (first function argument), decrements LDR_DDAG_NODE.LoadCount to zero, and calls RtlFreeHeap to deallocate LDR_DDAG_NODE DAG nodes. After LdrpMergeNodes returns, LdrpCondenseGraphRecurse calls LdrpDestroyNode to deallocate any DAG nodes in the LDR_DDAG_NODE.ServiceTagList list of the parent LDR_DDAG_NODE then deallocate the parent LDR_DDAG_NODE itself. LdrpCondenseGraphRecurse sets the state to LdrModulesCondensed before returning. Note: The LdrpCondenseGraphRecurse function and its callees rely heavily on all members of the LDR_DDAG_NODE structure, which needs further reverse engineering to fully understand the inner workings and "whys" of what's occurring here. Condensing is the process of discarding unnecessary nodes from the dependency graph.
LdrModulesReadyToInit (7)	LdrpNotifyLoadOfGraph	This state is set immediately before this function calls LdrpSendPostSnapNotifications to run post-snap DLL notification callbacks.
LdrModulesInitializing (8)	LdrpInitializeNode	Set at the start of this function, immediately before linking a module into the InInitializationOrderModuleList list. Loader lock (LdrpLoaderLock) is held here. Initializing is the process of running a module's initialization function (i.e. DllMain).
LdrModulesReadyToRun (9)	LdrpInitializeNode	Set at the end of this function, before it returns. Loader lock (LdrpLoaderLock) is held here.

See what a LDR_DDAG_STATE trace log looks like (be aware of the warnings).

Symbol/Export Lookup Comparison (Windows GetProcAddress vs POSIX dlsym GNU Implementation)

The Windows GetProcAddress and POSIX dlsym functions are platform equivalents because they resolve a symbol name to its address. They differ because GetProcAddress can only resolve function export symbols, whereas dlsym can resolve any external symbol. There's also this difference: The first argument of GetProcAddress requires passing in a module handle. In contrast, the first argument of dlsym can take a module handle, but it also accepts one of the RTLD_DEFAULT or RTLD_NEXT flags. Let's discuss how GetProcAddress functions first.

GetProcAddress receives an HMODULE (a module's base address) as its first argument. The loader maintains a red-black tree sorted by each module's base address called ntdll!LdrpModuleBaseAddressIndex. GetProcAddress -> LdrGetProcedureAddressForCaller -> LdrpFindLoadedDllByAddress (this is a call chain) searches this red-black tree for the matching module base address to ensure a valid DLL handle. Searching the LdrpModuleBaseAddressIndex red-black tree mandates acquiring the ntdll!LdrpModuleDataTableLock lock. If locating the module fails, GetProcAddress sets the thread error code (retrieved with the GetLastError function) to ERROR_MOD_NOT_FOUND. GetProcAddress receives a procedure name as a string for its second argument. GetProcAddress -> LdrGetProcedureAddressForCaller -> LdrpResolveProcedureAddress -> LdrpGetProcedureAddress calls RtlImageNtHeaderEx to get the NT header (IMAGE_NT_HEADERS) of the PE image. IMAGE_NT_HEADERS contains optional headers (IMAGE_OPTIONAL_HEADER), including the image data directory (IMAGE_DATA_DIRECTORY, this is in the .rdata section). The data directory (IMAGE_DATA_DIRECTORY) includes multiple directory entries, including IMAGE_DIRECTORY_ENTRY_EXPORT, IMAGE_DIRECTORY_ENTRY_IMPORT, IMAGE_DIRECTORY_ENTRY_RESOURCE, and more. LdrpGetProcedureAddress gets the PE's export directory entry. The compiler sorted the PE export directory entries alphabetically by procedure name ahead of time. LdrpGetProcedureAddress performs a binary search over the sorted export entry names, looking for a name matching the procedure name passed into GetProcAddress. If locating the procedure fails, GetProcAddress sets the thread error code to ERROR_PROC_NOT_FOUND. Locking isn't required while searching for an export because PE image exports are resolved once during library load and remain unchanged (this doesn't cover delay loading). In classic Windows monolithic fashion, GetProcAddress may do much more than what's stated here on edge cases. Still, for the sake of our comparison, we only need to know how GetProcAddress works at its core.

GNU's POSIX-compliant dlopen implementation firstly differs from GetProcAddress because the former will not validate a correct module handle before searching for a symbol. Pass in an invalid module handle, and the program will crash; to be fair, you deserve to crash if you do that. Also, not validating the module handle provides a great performance boost. Depending on the handle passed to dlopen, the GNU loader searches for symbols in a few ways. Here, we will cover the most straightforward case when calling dlsym with a handle to a library (i.e. dlsym("mylib.so", "myfunc")). The ELF standard specifies the use of a hash table for searching symbols. do_sym (called by _dl_sym) calls _dl_lookup_symbol_x to find the symbol in our specified library (also called an object). _dl_lookup_symbol_x calls _dl_new_hash to hash our symbol name with the djb2 hash function (look for the magic numbers). Recent versions of the GNU loader use this djb2 based hash function, which differs from the old standard ELF hash function based on the PJW hash function. At its introduction, this new hash function improved dynamic linking time by 50%. Though commonly used across Unix systems, this hash function is only a GNU extension and requires standardization. It's also worth noting that, in 2023, someone caught an overflow bug in the original hash function described by the System V Application Binary Interface. _dl_lookup_symbol_x calls do_lookup_x, where the real searching begins. do_lookup_x filters on our library's link_map to see if it should disregard searching it for any reason. Passing that check, do_lookup_x gets pointers into our library's DT_SYMTAB and DT_STRTAB ELF tables (the latter for use later while searching for matching symbol names). Based on our symbol's hash (calculated in _dl_new_hash), do_lookup_x selects a bucket from the hash table to search for symbols from. l_gnu_buckets is an array of buckets in our hash table to chose from. At build time during the linking phase, the linker builds each ELF image's hash table with the number of buckets, which adjusts depending on how many symbols are in the binary. With a bucket selected, do_lookup_x fetches the l_gnu_chain_zero chain for the given bucket and puts a reference to the chain in the hasharr pointer variable for easy access. l_gnu_chain_zero is an array of GNU hash entries containing standard ELF symbol indexes. The ELF symbol indexes inside are what's relevant to us right now. Each of these indexes points to a symbol table entry in the DT_SYMTAB table. do_lookup_x iterates through the symbol table entries in the chain until the selected symbol table entry holds the desired name or hits STN_UNDEF (i.e. 0), marking the end of the array. l_gnu_buckets and l_gnu_chain_zero inherit similarities in structure from the original ELF standardized l_buckets and l_chain arrays which the GNU loader still implements for backward compatibility. For the memory layout of the symbol hash table, see this diagram. There appears to be a fast path to quickly eliminate any entries that we can know early on won't match—passing the fast path check, do_lookup_x calls ELF_MACHINE_HASH_SYMIDX to extract the offset to the standard ELF symbol table index from within the GNU hash entry. The GNU hash entry is a layer on top of the standard ELF symbol table entry; in the old but standard hash table implementation, you can see that the chain is directly an array of symbol table indexes. Having obtained the symbol table index, dl_lookup_x passes the address to DT_SYMTAB at the offset of our symbol table index to the check_match function. The check_match function then examines the symbol name cross-referencing with DT_STRTAB where the ELF binary stores strings to see if we've found a matching symbol. Upon finding a symbol, check_match looks if the symbol requires a certain version (i.e. dlvsym GNU extension). In the unversioned case with dlsym, check_match determines if this is a hidden symbol (e.g. -fvisibility=hidden compiler option in GCC/Clang), and if so not returning the symbol. This loop restarts at the fast path check in dl_lookup_x until check_match finds a matching symbol or there are no more chain entries to search. Having found a matching symbol, do_lookup_symbol_x determines the symbol type; in this case, it's STB_GLOBAL and returns the successfully found symbol address. Finally, the loader internals pass this value back through the call chain until dlsym returns the symbol address to the user!

Note that the glibc source code loosely follows a Hungarian notation that prefixes structure members with l_ and structures with r_ (e.g. throughout the link_map structure).

Also, note that, unlike the Windows ntdll!LdrpHashTable (which serves an entirely different purpose), the hash table in each ELF DT_SYMTAB is made up of arrays instead of linked lists for each chain (each bucket has a chain). Using arrays (size determined during binary compilation) is possible because the ELF images aren't dynamically allocated structures like the LDR_DATA_TABLE_ENTRY structures ntdll!LdrpHashTable keeps track of. Arrays are significantly faster than linked lists because following links is an expensive operation (e.g. you lose all locality of reference).

Due to the increased locality of reference and a hash table being O(1) average and amortized time complexity vs a binary search being O(log n) time complexity, I believe that searching a hash table (bucket count optimized at compile time) and then iterating through the also optimally sized array is faster than the binary search approach employed by LdrpGetProcedureAddress in Windows for finding symbol addresses.

How Does GetProcAddress/dlsym Handle Concurrent Library Unload?

The dlsym function can locate RTLD_LOCAL symbols. Internally, the dlsym_implementation function acquires _dl_load_lock at its entry. Since dlclose must also acquire _dl_load_lock to unload a library, this prevents a library from unloading while another thread searches for a symbol in that same (or any) library. dlsym eventually calls into _dl_lookup_symbol_x to perform the symbol lookup. This coarse-grained locking approach is the only place the Windows loader approach is superior for maximizing concurrency and preventing deadlocks. I recommend that glibc switches use a more fine-grained locking approach like they already do with RTLD_GLOBAL symbols and their global scope locking system (GSCOPE). Or, try the Windows-like reference counting approach (although, this would require dlsym internally doing library unload if the reference count conurrently drops to zero thus acquiring _dl_load_lock anyway). Note that acquiring _dl_load_lock also allows dlsym to safely search link maps in the RTLD_DEFAULT and RTLD_NEXT pseudo-handle scenarios without acquiring _dl_load_write_lock. However, that goal could also be accomplished by shortly acquiring the _dl_load_write_lock lock in the _dl_find_dso_for_object function, so this is only a helpful side effect.

For GetProcAddress, LdrGetProcedureAddressForCaller (this is the NTDLL function GetProcAddress calls through to) acquires the LdrpModuleDatatableLock lock, searches for our module LDR_DATA_TABLE_ENTRY structure in the ntdll!LdrpModuleBaseAddressIndex red-black tree, checks if our DLL was loaded by dynamically loaded DLL (dynamically linked DLLs are already pinned by default, so they will never unload) if so atomically incrementing LDR_DATA_TABLE_ENTRY.ReferenceCount. LdrGetProcedureAddressForCaller releases the LdrpModuleDatatableLock lock. Before acquiring the LdrpModuleDatatableLock lock, note there's a special path for NTDLL whereby LdrGetProcedureAddressForCaller checks if the passed base address matches ntdll!LdrpSystemDllBase (this holds NTDLL's base address) and lets it skip close to the meat of the function where LdrpFindLoadedDllByAddress (note: searching for the module occurs twice for any other module. why?) and LdrpResolveProcedureAddress occurs. Towards the end of LdrGetProcedureAddressForCaller, it calls LdrpDereferenceModule, passing the LDR_DATA_TABLE_ENTRY of the module it was searching for a procedure in. LdrpDereferenceModule, assuming the module isn't pinned (LDR_ADDREF_DLL_PIN) or a static import (ProcessStaticImport in LDR_DATA_TABLE_ENTRY.Flags), atomically decrements the same LDR_DATA_TABLE_ENTRY.ReferenceCount of the searched module. Note that the LdrpDerferenceModule function doesn't perform module deinitialization (DLL_PROCESS_DETACH; requiring loader lock) before unmapping the module, so I can only assume that deinitialization doesn't happen in this case (unless I'm missing something but I don't think I am). By proper locking and reference counting, GetProcAddress ensures a module isn't unmapped midway through its search.

Note that the above analysis of Windows GetProcAddress internals is only a rough look-through.

In any case, if you're code frees libraries whose exports/symbols are still in use (after locating them with GetProcAddress or dlsym), then you can expect your application to crash due to your own negligence.

GNU Loader Lock Hierarchy

_dl_load_lock is higher than _dl_load_write_lock in the lock hierarchy. There are occurrences where the GNU loader walks the list of link maps (a read-only operation) while only holding the _dl_load_lock lock. For instance, walking the link map list while only holding _dl_load_lock but not dl_load_write_lock occurs in _dl_sym_find_caller_link_map -> _dl_find_dso_for_object. Linking a new link map into the list of link maps (a write operation) requires acquiring _dl_load_lock then dl_load_write_lock (e.g. see load-library experiment). It's unsafe to modify the link map list without acquiring _dl_load_write_lock because, for instance, the __dl_iterate_phdr function acquires _dl_load_write_lock without first acquiring _dl_load_lock to ensure consistency of the link map list.

This lock hierarchy differs from the Windows LdrpLoaderLock and LdrpModuleDataTableLock locks, where no hierarchy exists between them.

The _dl_load_lock lock is the highest in the lock hierarchy and is also above GSCOPE locking.

Lazy Linking Synchronization

The above section only covers locking for the GetProcAddress/dlsym functions, which resolve a symbol name to an address. Lazy linking also does this; however, the linker using statically allocated memory in the PE/ELF images and more known variables due to offloading some upfront front work to process startup (e.g. no need to search for a module) makes concurrency easier.

Stepping into a lazy linked function on the GNU loader reveals that it eventually calls the familiar _dl_lookup_symbol_x function to locate a symbol. During dynamic linking, _dl_lookup_symbol_x can resolve global symbols. The GNU loader uses GSCOPE, the global scope system, to ensure consistent access to STB_GLOBAL symbols (as it's known in the ELF standard; this maps to the RTLD_GLOBAL flag of POSIX dlopen). The global scope is the searchlist in the main link map. The main link map refers to the program's link map (not one of the libraries). In the TCB (this is the generic term for the Windows TEB) of each thread is a piece of boolean atomic state (this is not a reference count), which may also hold a flag known as the gscope_flag that keeps track of which threads are depending on the global scope for their operations. A thread uses the THREAD_GSCOPE_SET_FLAG macro (internally calls THREAD_SETMEM) to atomically set this flag and the THREAD_GSCOPE_RESET_FLAG macro to atomically unset this flag. When the GNU loader requires synchronization of the global scope, it uses the THREAD_GSCOPE_WAIT macro to call __thread_gscope_wait. Note there are two implementations of __thread_gscope_wait, one for the Native POSIX Threads Library (NPTL) used on Linux systems and the other for the Hurd Threads Library (HTL) which was previously used on GNU Hurd systems (with the GNU Mach microkernel). GNU Hurd has since switched to using NPTL. For our purposes, only NPTL is relevant. The __thread_gscope_wait function iterates through the gscope_flag of all (user and system) threads, signalling to them that it's waiting (THREAD_GSCOPE_FLAG_WAIT) to synchronize. GSCOPE is a unique approach to locking that works by the GNU loader creating low-level locking primitives. This synchronization technique can similarly be done on Windows using the Wait­On­Address and Wake­By­Address­Single/Wake­By­Address­All functions. Note that the THREAD_SETMEM and THREAD_GSCOPE_RESET_FLAG macros don't prepend a lock prefix to the assembly instruction when atomically modifying a thread's gscope_flag. These modifications are still atomic because xchg is atomic by default on x86, and a single aligned load or store (e.g. mov) operation is also atomic by default on x86 up to 64 bits. If gscope_flag were a reference count, the assembly instruction would require a lock prefix (e.g. lock inc) because incrementing requires two operations, one load and one store. The GNU loader must still use a locking modification on architectures where memory consistency doesn't automatically guarantee this (e.g. AArch64). Also, note that all assembly in the glibc source code is in AT&T syntax, not Intel syntax.

The Windows loader resolves lazy linkage by calling LdrResolveDelayLoadedAPI (an NTDLL export). This function makes heavy use of LDR_DATA_TABLE_ENTRY.LoadContext. Symbols for the LoadContext structure aren't publicly available, and some members require reverse engineering. In the protected load case, LdrpHandleProtectedDelayload -> LdrpWriteBackProtectedDelayLoad (LdrResolveDelayLoadedAPI may call this or LdrpHandleUnprotectedDelayLoad) acquires the LDR_DATA_TABLE_ENTRY.Lock lock and never calls through to LdrpResolveProcedureAddress. The unprotected case eventually calls through to the same LdrpResolveProcedureAddress function that GetProcAddress uses; this is not true for the protected case. From glazing over some of the functions LdrResolveDelayLoadedAPI calls, a module's LDR_DATA_TABLE_ENTRY.ReferenceCount is atomically modified quite a bit. Setting a breakpoint on both LdrpHandleProtectedDelayload and LdrpHandleUnprotectedDelayLoad shows that a typical process will only ever call the former function (experimented with a ShellExecute).

The above paragraph is only an informal glance; I or someone else must thoroughly look into delay loading in the modern Windows loader.

Loader Enclaves

An enclave is a security feature that isolates a region of data or code within an application's address space. Enclaves utilize one of three backing technologies to provide this security feature: Intel Software Guard Extensions (SGX), AMD Secure Encrypted Virtualization (SEV), or Virtualization-based Security (VBS). The Intel and AMD solutions are memory encryptors; they safeguard sensitive memory by encrypting it at the hardware level. VBS securely isolates sensitive memory by containing it in virtual secure mode where even the NT kernel cannot access it.

Within SGX, there is SGX1 and SGX2. SGX1 only allows using statically allocated memory with a set size before enclave initialization. SGX2 adds support for an enclave to allocate memory dynamically.

Due to this limitation, putting a library into an enclave on SGX1 requires that the library be statically linked. On the other hand, SGX2 supports dynamically loading/linking libraries.

A Windows VBS-based enclave requires Microsoft's signature as the root in the chain of trust or as a countersignature on a third party's certificate of an Authenticode-signed DLL. The signature must contain a specific extended key usage (EKU) value that permits running as an enclave. Enabling test signing on your system can get an unsigned enclave running. VBS enclaves may call enclave-compatible versions of Windows API functions from the enclave version of the same library.

Both Windows and Linux support loading libraries into enclaves. An enclave library requires special preparation; a programmer cannot load any generic library into an enclave.

Windows integrates enclaves as part of their loader. One may CreateEnclave to make an enclave and then call LoadEnclaveImage to load a library into an enclave. Internally, CreateEnclave (public API) calls ntdll!LdrCreateEnclave, which then calls ntdll!NtCreateEnclave; this function only does the system call to create an enclave, and then calls LdrpCreateSoftwareEnclave to initialize and link the new enclave entry into the ntdll!LdrpEnclaveList list. ntdll!LdrpEnclaveList is the list head, its list entries of type LDR_SOFTWARE_ENCLAVE are allocated onto the heap and linked into the list. Compiling an enclave library requires special compilation steps (e.g. compiling for Intel SGX).

The GNU loader has no knowledge of enclaves. Intel provides the Intel SGX SDK and the Intel SGX Platform Software (PSW) necessary for using SGX on Linux. The SGX driver awaits upstreaming into the Linux source tree. Linux currently has no equivalent to Windows Virtualization Based Security. Here's some sample code for loading an SGX module. However, Hypervisor-Enforced Kernel Integrity (Heki) is on its way, including patches to the kernel. Developers at Microsoft are introducing this new feature into Linux.

If you're interested, here's a comparison of Intel SGX and AMD SEV.

Open Enclave is a cross-platform and hardware-agnostic open source library for utilizing enclaves. Here's some sample code calling into an enclave library.

The Windows LdrpObtainLockedEnclave Function

The Windows loader uses ntdll!LdrpObtainLockedEnclave to obtain the enclave lock for a module, doing per-node locking. The LdrpObtainLockedEnclave function acquires the ntdll!LdrpEnclaveListLock lock and then searches from the ntdll!LdrpEnclaveList list head to find an enclave for the DLL image base address this function receives as its first argument. If LdrpObtainLockedEnclave finds a matching enclave, it atomically increments a reference count and enters a critical section stored in the enclave structure before returning that enclave's address to the caller. Typically (unless your process uses enclaves), the list at ntdll!LdrpEnclaveList will be empty, effectively making LdrpObtainLockedEnclave a no-op.

The LdrpObtainLockedEnclave function came up in LdrGetProcedureAddressForCaller (called by GetProcAddress), so I wanted to give it a look.

Modern Windows Loader Synchronization: State > Locks

Me from the future: This section needs to be split up and revised. All of the facts and analysis are correct, but the thesis is weak. For instance, ntdll!LdrpWorkInProgress is protected by ntdll!LdrpWorkQueueLock; however, you can't know the state of ntdll!LdrpWorkInProgress by testing that lock because the loader also uses that lock to protect the work queue. It's the same thing with loader lock in the following case study due to process exit. In trying to support my thesis, I actually found evidence disproving it.

In the modern Windows loader, there's a clear trend regarding how the Windows loader performs synchronization: State is more important than locks. Specifically, this state is more globalized than it is in a legacy Windows loader, which is likely, in large part, to support the parallel loader.

The loader primarily uses locks to protect the state. State, kept consistent through the use of locks, is what the loader predominantly uses to decide what actions it should perform. So, there's a level of indirection here.

Some good examples of such state include (see the relevant locks): LDR_DDAG_NODE.LDR_DDAG_STATE, ntdll!LdrpWorkInProgress, and LoadOwner in TEB.SameTebFlags)

There's one minor exception to this rule where the loader regularly depends on the locking status of a lock itself (minor because this lock isn't central to the loader state machine, and a typical process won't even run any DLL notification callbacks): ntdll!LdrpDllNotificationLock

In contrast, a legacy Windows loader relies solely on locks (predominantly just loader lock), locking and unlocking, and some local state (often passed as a function argument). Sometimes, also checking the state of loader lock directly to determine control flow.

Search Wine source code: RtlIsCriticalSectionLockedByThread "LoaderLock"

Case Study

Let me show you what I mean because comparing the modern LdrGetProcedureAddressForCaller with the old (ReactOS) equivalent function of LdrpGetProcedureAddress is a textbook example of this. GetProcAddress is the public Windows API function that calls through to these NTDLL exports.

Internally, when GetProcAddress resolves an exported procedure to an address, it checks if the loader has performed initialization on the containing module. If the module requires initialization, then GetProcAddress initializes the module internally (i.e. calling its DllMain) before returning a procedure address to the caller.

In the ReactOS code for LdrpGetProcedureAddress, we see this happen:

    /* Acquire lock unless we are initing */
    /* MY COMMENT: This refers to the loader initing; ignore this for now */
    if (!LdrpInLdrInit) RtlEnterCriticalSection(&LdrpLoaderLock);

    ...

        /* Finally, see if we're supposed to run the init routines */
        /* MY COMMENT: ExecuteInit is a function argument which GetProcAddress always passes as true */
        if ((NT_SUCCESS(Status)) && (ExecuteInit))
        {
            /*
            * It's possible a forwarded entry had us load the DLL. In that case,
            * then we will call its DllMain. Use the last loaded DLL for this.
            */
            Entry = NtCurrentPeb()->Ldr->InInitializationOrderModuleList.Blink;
            LdrEntry = CONTAINING_RECORD(Entry,
                                         LDR_DATA_TABLE_ENTRY,
                                         InInitializationOrderLinks);
 
            /* Make sure we didn't process it yet*/
            /* MY COMMENT: If module initialization hasn't already run... */
            if (!(LdrEntry->Flags & LDRP_ENTRY_PROCESSED))
            {
                /* Call the init routine */
                _SEH2_TRY
                {
                    Status = LdrpRunInitializeRoutines(NULL);
                }
                _SEH2_EXCEPT(EXCEPTION_EXECUTE_HANDLER)
                {
                    /* Get the exception code */
                    Status = _SEH2_GetExceptionCode();
                }
                _SEH2_END;
            }
        }

...

This code is perfectly safe, as we know that even the legacy loader uses loader lock to protect against concurrent initialization/deinitialization or load/unload (amongst other things).

Now, let's see how the modern Windows loader handles this. In LdrGetProcedureAddressForCaller, there's an instance where module initialization may occur without LdrGetProcedureAddressForCaller itself acquiring loader lock (what follows is a marked-up IDA decompilation):

...
    ReturnValue = LdrpResolveProcedureAddress(
                    (unsigned int)v24,
                    (unsigned int)Current_LDR_DATA_TABLE_ENTRY,
                    (unsigned int)Heap,
                    v38,
                    v30,
                    (char **)&v34);
...
    // If LdrpResolveProcedureAddress succeeds
    if ( ReturnValue >= 0 )
    {
      // Test for the searched module having a LDR_DDAG_STATE of LdrModulesReadyToInit 
      if ( Current_Module_LDR_DDAG_STATE == 7
        // Test whether module module initialization should occur
        // LdrGetProcedureAddressForCaller receives this as its fifth argument, KERNELBASE!GetProcAddresForCaller (called by KERNELBASE!GetProcAddress) always sets it to TRUE
        && ExecuteInit
        // Test for the current thread having the LoadOwner flag in TEB.SameTebFlags
        && (NtCurrentTeb()->SameTebFlags & 0x1000) != 0
        // Test for the current thread not holding the LdrpDllNotificationLock lock (i.e. we're not executing a DLL notification callback)
        && !(unsigned int)RtlIsCriticalSectionLockedByThread(&LdrpDllNotificationLock) )
      {
        DdagNode = *(_QWORD *)(Current_LDR_DATA_TABLE_ENTRY + 0x98);
        v33[0] = 0;
        // Perform module initialization
        ReturnValue = LdrpInitializeGraphRecurse(DdagNode, 0i64, v33);
      }
...

Huh? LdrGetProcedureAddressForCaller didn't acquire loader lock, yet it is performing module initialization! How could that be safe?

It's a layer of abstraction. By testing the module's LDR_DDAG_STATE, LoadOwner in TEB.SameTebFlags, ntdll!LdrpDllNotificationLock, and ExecuteInit states, LdrGetProcedureAddressForCaller knows that the only way the loader would have let it reach this point where all these conditions pass is if it were safe to do module initialization. LdrGetProcedureAddressForCaller doesn't have to be aware of loader lock because it trusts the loader state machine to have done the right thing before handing off execution to LdrGetProcedureAddressForCaller.

This premise holds up when considering where the loader must be for all conditions to pass. Two of the things the LdrpPrepareModuleForExecution function does is run DLL notification callbacks (calls LdrpNotifyLoadOfGraph) and then perform module initialization (calls LdrpInitializeGraphRecurse). We know a module's LDR_DDAG_STATE is set to LdrModulesReadyToInit immediately before running any DLL notification callbacks. DLL notification callbacks may call into third-party code (anyone can register for a callback), and a thread executing a DLL notification callback holds ntdll!LdrpDllNotificationLock as it runs. LoadOwner in TEB.SameTebFlags correlates with ntdll!LdrpWorkInProgress; it's what a thread uses to inform itself that it's responsible for the work in progress. Lastly, ExecuteInit is the fifth argument passed to LdrGetProcedureAddressForCaller, so should NTDLL want to get the address of a procedure within a module while the conditions up to this point pass without doing module initialization, it could simply pass the ExecuteInit argument as FALSE. By process of elimination, this logically leaves module initialization (i.e. DllMain) as the only place this code path of LdrGetProcedureAddressForCaller could execute. And everybody knows that DllMain is run under loader lock.

Therefore, by testing all these conditions, LdrGetProcedureAddressForCaller knows that the loader will already have acquired loader lock if need be. Having passed these conditions, LdrGetProcedureAddressForCaller could also recursively acquire loader lock itself. However, there's no point in expending the extra resources to do that when it's already known to be safe to perform module initialization now. Additionally, other than module initialization, a process acquires loader lock at its exit (RtlExitUserProcess), so testing these conditions instead of loader lock ensures LdrGetProcedureAddressForCaller doesn't initialize a module during this phase of the process.

Continue to the next section for information regarding that "if need be" part in the context of the wider loader.

Windows Loader Initialization Locking Requirements

Reading the ReactOS LdrGetProcedureAddressForCaller code, you saw this line of code:

/* Acquire lock unless we are initing */
/* MY COMMENT: This refers to the loader initing */
if (!LdrpInLdrInit) RtlEnterCriticalSection(&LdrpLoaderLock);

The loader won't acquire loader lock during module initialization when the loader is initializing (at process startup). What's up with that? It's a performance optimization to forego locking during process startup.

Not acquiring loader lock here is safe because the legacy loader, like the modern loader, includes a mechanism for blocking new threads spawned into the process until loader initialization is complete. The legacy loader waits in the LdrpInit function using a ntdll!LdrpProcessInitialized spinlock and sleeping with ZwDelayExecution/NtDelayExecution. The LdrpInit function sets LdrpInLdrInit to FALSE, initializes the process by calling LdrpInitializeProcess (LdrpInitializeProcess includes initializing all with DLLs with the LdrpCallInitRoutine similar to how the modern loader does with the LdrpInitializeGraphRecurse function), then upon returning, LdrpInit sets LdrpInLdrInit to TRUE. Hence, during LdrpInLdrInit, one can safely forgo acquiring loader lock.

The modern loader optimizes waiting by using the LdrpInitCompleteEvent Win32 event instead of sleeping for a set time. The modern loader also includes a ntdll!LdrpProcessInitialized spinlock. However, the loader may only spin on it until event creation (NtCreateEvent), at which point the loader waits solely using that synchronization object. While ntdll!LdrInitState is 0, it's safe not to acquire any locks. This includes accessing shared module information data structures without acquiring ntdll!LdrpModuleDataTableLock and performing module initialization/deinitialization. ntdll!LdrInitState changes to 1 immediately after LdrpInitializeProcess calls LdrpEnableParallelLoading which creates the loader work threads. However, these loader work threads won't have any work yet, so it should still be safe not to acquire locks during this time. In practice, one can see that the loader may call the LdrpInitShimEngine function before LdrpEnableParallelLoading (not always; it happens when running Google Chrome under WinDbg). The LdrpInitShimEngine function calls LdrpLoadShimEngine, which does a whole bunch of typically unsafe actions like module initialization (calls LdrpInitalizeNode directly and calls LdrpInitializeShimDllDependencies, which in turn calls LdrpInitializeGraphRecurse) without loader lock and walking the PEB_LDR_DATA.InLoadOrderModuleList without acquiring ntdll!LdrpModuleDataTableLock. Of course, all these actions are safe due to the unique circumstances of process initialization. Note that the shim initialization function may still acquire the LdrpDllNotificationLock lock not for safety but because the loader branches on its state using the RtlIsCriticalSectionLockedByThread function.

The modern loader explicitly checks ntdll!LdrInitState to optionally perform locking as an optimization in a few places. Notably, the ntdll!RtlpxLookupFunctionTable function (this is in connection to the ntdll!LdrpInvertedFunctionTable data structure, which is something I haven't looked into yet) opts to perform no locking (ntdll!LdrpInvertedFunctionTableSRWLock) before accessing shared data structures if ntdll!LdrInitState does not equal 3 (i.e. loader initialization is done). Similarly, the ntdll!LdrLockLoaderLock function only acquires loader lock if loader initialization is done.

The GNU loader doesn't implement any such performance hack to forego locking on process startup. The absence of any such mechanism by the GNU loader enables threads to start and exit from within a constructor at process startup. Therefore, the GNU loader is more flexible in this respect.

ELF Flat Symbol Namespace (GNU Namespaces and STB_GNU_UNIQUE)

Windows EXE and MacOS Mach-O starting with OS X v10.1 executable formats store library symbols in a two-dimensional namespace; thus effectively making every library its own namespace.

On the other hand, the Linux ELF executable format specifies a flat namespace such that two libraries having the same symbol can collide. For example, if two libraries expose a malloc function, then the dynamic linker won't be able to differentiate between them. As a result, the dynamic linker recognizes the first malloc symbol definition it sees as the malloc, ignoring any malloc definitions that come later. This refers to cases of loading a library (dlopen) with RTLD_GLOBAL; with RTLD_LOCAL, the newly loaded library's symbols aren't made available to other libraries in the process.

These namespace collisions have been the source of some bugs, and as a result, there have been workarounds to fix them. The most straightforward being: dlsym("MyLibrary.so", "malloc"). However, there are plenty of cases where that doesn't cut it, so GNU devised some solutions of their own.

Let's talk about GNU loader namespaces. Since 2004, Glibc loader has supported a feature known as loader namespaces for separating symbols contained within a loaded library into a separate namespace. Creating a new namespace for a loading library requires calling dlmopen (this is a GNU extension). Loader namespaces allow a programmer to isolate a module to its own namespace. There are various reasons GNU gives for why a developer might want to do this and why RTLD_LOCAL isn't a substitute for this, mainly regarding loading an untrusted library with an unwieldy number of generic symbol names polluting your namespace. It's worth noting that there's a hard limit of 16 on the number of GNU loader namespaces a process can have. Some might say this is just a bandage patch around the core issue of flat ELF symbol namespaces, but it doesn't hurt to exist.

In 2009, the GNU loader received a new symbol type calledSTB_GNU_UNIQUE. In the dl_lookup_x function, determining the symbol type is the final step after successfully locating a symbol. We previously saw this occur when our symbol was determined to be type STB_GLOBAL. STB_GNU_UNIQUE is another one of those symbol types, and as the name implies, it's a GNU extension. The purpose of this new symbol type was to fix symbol collision problems in C++ by giving precedence to STB_GNU_UNIQUE symbols even above RTLD_LOCAL symbols within the same module. Binding symbols with the STB_GNU_UNIQUE symbol type became how GCC compiles C++ binaries. Due to the global nature of STB_GNU_UNIQUE symbols and their lack of a reference counting feature (this could impact performance), their usage in a library makes unloading impossible by design. On top of this, STB_GNU_UNIQUE introduced significant bloat to the GNU loader by adding a special, dynamically allocated hash table known as the _ns_unique_sym_table just for handling this one new symbol type, which, along with a lock for controlling access to this new data structure, is included in the global _rtld_global. rtld_global (run-time dynamic linker) being the structure that defines all variables global to the loader. There's also a performance hit because locating a STB_GNU_UNIQUE symbol requires a lookup in two separate hash tables.

These workarounds prove that ELF symbol namespaces should have been two-dimensional long ago. Standards organizations could still modify the ELF standard to support two-dimensional namespaces (although it would likely require a new version or an ABI compatibility hack); that's what Apple did with Mac OS a long time ago.

What's the Difference Between Dynamic Linking, Static Linking, Dynamic Loading, Static Loading?

Loading and linking are two operations essential for program execution. A loader is responsible for placing executable modules into the program's memory for execution. Linking resolves dependencies between executable modules or stitches executable modules together to create a runnable program.

Dynamic linking resolves dependencies between executable modules. Dynamic linking typically occurs when a program load/run time; however, it can also occur later due to a lazy linking optimization.

Static linking is when a compiler (e.g. GCC, Clang, and MSVC) uses a linker (e.g. the ld program on GNU/Linux or link.exe program on MSVC) to stitch object files together into single executables. Linkers can also write information into a program about where a dynamic linker or loader can find its dependencies at process load/run time.

Dynamic loading refers to loading a library at run-time with dlopen/LoadLibrary.

Static loading refers to loading that occurs due to dependencies found during dynamic linking.

LoadLibrary vs dlopen Return Type

On Windows, LoadLibrary returns an officially opaque HMODULE, which is implemented as the base address of the loaded module.

In POSIX, dlopen returns a symbol table handle. On my GNU/Linux system, this handle points to the object's own link_map structure located in the heap. This handle is opaque, meaning you must not access the contents behind it directly (they could change between versions, and it's implementation dependent); instead, only pass this handle to other dl* functions.

How cool is that? That's like if Windows serviced your LoadLibrary request by handing you back the module's LDR_DATA_TABLE_ENTRY!

DllMain (Loader Lock) + COM and CLR

Trying to create a COM server under loader lock fails deterministically. For instance, this code run from the DllMain on DLL_PROCESS_ATTACH will deadlock:

HRESULT hres;
HWND hwnd = GetDesktopWindow();
// Ensure valid LNK file with this CMD command:
// explorer "C:\ProgramData\Microsoft\Windows\Start Menu\Programs\Accessories\Notepad.lnk"
LPCSTR linkFilePath = "C:\\ProgramData\\Microsoft\\Windows\\Start Menu\\Programs\\Accessories\\Notepad.lnk";
WCHAR resolvedPath[MAX_PATH];
       
hres = CoInitializeEx(NULL, 0);

if (SUCCEEDED(hres)) {
    // Implementation: https://learn.microsoft.com/en-us/windows/win32/shell/links#resolving-a-shortcut
    ResolveIt(hwnd, linkFilePath, resolvedPath, MAX_PATH);
}

CoUninitialize();

Deadlocked call stack:

0:000> k
 # Child-SP          RetAddr               Call Site
00 00000080`f96fcde8 00007ffd`26c93f8f     ntdll!NtAlpcSendWaitReceivePort+0x14
01 00000080`f96fcdf0 00007ffd`26ca94d7     RPCRT4!LRPC_BASE_CCALL::SendReceive+0x12f
02 00000080`f96fcec0 00007ffd`26c517c0     RPCRT4!NdrpSendReceive+0x97
03 00000080`f96fcef0 00007ffd`26c524bf     RPCRT4!NdrpClientCall2+0x5d0
04 00000080`f96fd510 00007ffd`28491ce5     RPCRT4!NdrClientCall2+0x1f
05 (Inline Function) --------`--------     combase!ServerAllocateOXIDAndOIDs+0x73 [onecore\com\combase\idl\internal\daytona\objfre\amd64\lclor_c.c @ 313] 
06 00000080`f96fd540 00007ffd`28491acd     combase!CRpcResolver::ServerRegisterOXID+0xd5 [onecore\com\combase\dcomrem\resolver.cxx @ 1056] 
07 00000080`f96fd600 00007ffd`28494531     combase!OXIDEntry::RegisterOXIDAndOIDs+0x71 [onecore\com\combase\dcomrem\ipidtbl.cxx @ 1642] 
08 (Inline Function) --------`--------     combase!OXIDEntry::AllocOIDs+0xc2 [onecore\com\combase\dcomrem\ipidtbl.cxx @ 1696] 
09 00000080`f96fd710 00007ffd`2849438f     combase!CComApartment::CallTheResolver+0x14d [onecore\com\combase\dcomrem\aprtmnt.cxx @ 693] 
0a 00000080`f96fd8c0 00007ffd`284abc2f     combase!CComApartment::InitRemoting+0x25b [onecore\com\combase\dcomrem\aprtmnt.cxx @ 991] 
0b (Inline Function) --------`--------     combase!CComApartment::StartServer+0x52 [onecore\com\combase\dcomrem\aprtmnt.cxx @ 1214] 
0c 00000080`f96fd930 00007ffd`2849c285     combase!InitChannelIfNecessary+0xbf [onecore\com\combase\dcomrem\channelb.cxx @ 1028] 
0d 00000080`f96fd960 00007ffd`2849a644     combase!CGIPTable::RegisterInterfaceInGlobalHlp+0x61 [onecore\com\combase\dcomrem\giptbl.cxx @ 815] 
0e 00000080`f96fda10 00007ffd`21b86399     combase!CGIPTable::RegisterInterfaceInGlobal+0x14 [onecore\com\combase\dcomrem\giptbl.cxx @ 776] 
0f 00000080`f96fda50 00007ffd`21b5adb3     PROPSYS!CApartmentLocalObject::_RegisterInterfaceInGIT+0x81
10 00000080`f96fda90 00007ffd`21b842e6     PROPSYS!CApartmentLocalObject::_SetApartmentObject+0x7b
11 00000080`f96fdac0 00007ffd`21b5c1fc     PROPSYS!CApartmentLocalObject::TrySetApartmentObject+0x4e
12 00000080`f96fdaf0 00007ffd`21b5bde6     PROPSYS!CreateObjectWithCachedFactory+0x2bc
13 00000080`f96fdbd0 00007ffd`21b5d16c     PROPSYS!CreateMultiplexPropertyStore+0x46
14 00000080`f96fdc30 00007ffd`241d3235     PROPSYS!PSCreateItemStoresFromDelegate+0xbfc
15 00000080`f96fde90 00007ffd`2422892f     windows_storage!CShellItem::_GetPropertyStoreWorker+0x2d5
16 00000080`f96fe3d0 00007ffd`2422b7e7     windows_storage!CShellItem::GetPropertyStoreForKeys+0x14f
17 00000080`f96fe6a0 00007ffd`2415f2b6     windows_storage!CShellItem::GetCLSID+0x67
18 00000080`f96fe760 00007ffd`2415eb0b     windows_storage!GetParentNamespaceCLSID+0xde
19 00000080`f96fe7c0 00007ffd`241772fb     windows_storage!CShellLink::_LoadFromStream+0x2d3
1a 00000080`f96feaf0 00007ffd`2417709c     windows_storage!CShellLink::LoadFromPathHelper+0x97
1b 00000080`f96feb40 00007ffd`24177039     windows_storage!CShellLink::_LoadFromFile+0x48
1c 00000080`f96febd0 00007ffd`21aa10e2     windows_storage!CShellLink::Load+0x29
1d (Inline Function) --------`--------     TestDLL!ResolveIt+0x8c [C:\Users\user\source\repos\TestDLL\TestDLL\dllmain.cpp @ 110] 
1e 00000080`f96fec00 00007ffd`21aa143b     TestDLL!DllMain+0xd2 [C:\Users\user\source\repos\TestDLL\TestDLL\dllmain.cpp @ 170] 
1f 00000080`f96ff4f0 00007ffd`28929a1d     TestDLL!dllmain_dispatch+0x8f [d:\a01\_work\20\s\src\vctools\crt\vcstartup\src\startup\dll_dllmain.cpp @ 281] 
20 00000080`f96ff550 00007ffd`2897c2c7     ntdll!LdrpCallInitRoutine+0x61
21 00000080`f96ff5c0 00007ffd`2897c05a     ntdll!LdrpInitializeNode+0x1d3
22 00000080`f96ff710 00007ffd`2894d947     ntdll!LdrpInitializeGraphRecurse+0x42
23 00000080`f96ff750 00007ffd`2892fbae     ntdll!LdrpPrepareModuleForExecution+0xbf
24 00000080`f96ff790 00007ffd`289273e4     ntdll!LdrpLoadDllInternal+0x19a
25 00000080`f96ff810 00007ffd`28926af4     ntdll!LdrpLoadDll+0xa8
26 00000080`f96ff9c0 00007ffd`260156b2     ntdll!LdrLoadDll+0xe4
27 00000080`f96ffab0 00007ff7`8fda1022     KERNELBASE!LoadLibraryExW+0x162
28 00000080`f96ffb20 00007ff7`8fda1260     EmptyProject!main+0x12 [C:\Users\user\source\repos\EmptyProject\EmptyProject\source.c @ 82] 
29 (Inline Function) --------`--------     EmptyProject!invoke_main+0x22 [d:\a01\_work\20\s\src\vctools\crt\vcstartup\src\startup\exe_common.inl @ 78] 
2a 00000080`f96ffb50 00007ffd`26f37344     EmptyProject!__scrt_common_main_seh+0x10c [d:\a01\_work\20\s\src\vctools\crt\vcstartup\src\startup\exe_common.inl @ 288] 
2b 00000080`f96ffb90 00007ffd`289626b1     KERNEL32!BaseThreadInitThunk+0x14
2c 00000080`f96ffbc0 00000000`00000000     ntdll!RtlUserThreadStart+0x21

Notice that the deadlock occurs on the first call to a COM method (hres = ppf->Load(wsz, STGM_READ)), not when CoCreateInstance creates an in-process COM server (CLSCTX_INPROC_SERVER). The deadlock occurs here because a COM server's startup is lazy loading. Lazy loading occurs for the most common CLSCTX_INPROC_SERVER but not necessarily for other execution contexts such as CLSCTX_LOCAL_SERVER (thereby causing the deadlock within CoCreateInstance itself). Also, notice this call stack is similar to the deadlock we receive from ShellExecute under DllMain.

As we know from "Perfect DLL Hijacking", loader lock is the root cause of this deadlock and releasing this lock allows execution to continue (also, see the DEBUG NOTICE in the LdrLockLiberator project). However, setting a read watchpoint on loader lock reveals that user-mode COM internals never check the state of the loader lock. This finding leads me to believe that the remote procedure call by ntdll!NtAlpcSendWaitReceivePort causes a remote process (csrss.exe?) to introspect on the state of loader lock within our process, potentially explaining why WinDbg never hits our watchpoint. Starting a COM server (combase!CComApartment::StartServer) involves calling NdrClientCall2 to perform a remote procedure call. In particular, the combase!ServerAllocateOXIDAndOID function does RPC according to the lclor interface.

Microsoft explains this deadlock (emphasis mine):

To do it, the Windows loader uses a process-global critical section (often called the "loader lock") that prevents unsafe access during module initialization.

The CLR will attempt to automatically load that assembly, which may require the Windows loader to block on the loader lock. A deadlock occurs since the loader lock is already held by code earlier in the call sequence.

Note that Microsoft wrote this documentation to resolve loader lock issues high-level developers may come across with the CLR (the .NET runtime environment); however, it also roughly applies to COM (where the deadlock occurs in the call stack looks similar). I reason that initializing a COM object, like a CLR assembly, may take "actions that are invalid under loader". So, instead of allowing for non-determinism, Microsoft took steps to make starting an in-process COM server deadlock deterministically.

Emphasis on the word "block" because it indicates that, in this case, the process treats loader lock as an exclusive/write lock instead of a critical section (a thread synchronization mechanism), the latter of which allows recursive acquisition. This fact aligns with what we see when starting a COM server from DllMain. One might wonder why this RPC call has to block on loader lock instead of recursively acquiring it to initialize any uninitialized libraries (since this is all within the same thread and Microsoft built the loader module initialization functionality to be reentrant). After all, parts of the loader are effectively already doing this to ensure module initialization. Additionally, since loader lock is already acquired, there's no risk of lock hierarchy violation (potentially leading to an ABBA deadlock) by recursively acquiring it. The answer is to ensure the CLR or COM initialization/constructor functions run without loader lock. Ensuring this is necessary because CLR loader initialization functions execute under the assumption of no loader lock. Compiling a binary with the /clr option causes the MSVC compiler to put constructors in the non-standard .cctor section. The .cctor section is Microsoft's method for guaranteeing separation between the native and CLR loaders. COM doesn't have a separate high-level loader, so Microsoft ensures COM initialization/constructor functions are never run under loader lock by deterministically deadlocking.

Please note that the above analysis is a strong theory based on the surrounding evidence and my work. However, verifying it with full certainty would require checking whether the remote process is checking the loader lock of our process. I want to figure out how to debug this; please do contribute if you know.

In the documentation for CoInitializeEx lies this statement:

Because there is no way to control the order in which in-process servers are loaded or unloaded, do not call CoInitialize, CoInitializeEx, or CoUninitialize from the DllMain function.

This warning concerns a lock hierarchy violation risk between the native (e.g. ntdll!LdrpLoaderLock) and COM (e.g. combase!gComLock, combase!gChannelInitLock, and combase!gOXIDLock) synchronization mechanisms. Realizing this risk, however unlikely, would lead to an ABBA deadlock. Ideally, Microsoft would ensure a clean lock hierarchy between these subsystems by controlling load/unload order, adding some shared piece of state to synchronize on before acquiring locks, or otherwise (it's perhaps easier said than done given the monolithic nature of it all). Microsoft has realized this risk in their code (note that starting with OLE 2, OLE is built on top of COM, so they share internals). The simplest method for fixing this ABBA deadlock would be to acquire the loader lock (it's in the PEB) before the COM lock, thereby keeping a clean lock hierarchy. However, this method would decrease concurrency, thus reducing performance. Another solution would be to specifically target COM functions that interact with the native loader while already holding the COM lock, such as CoFreeUnused­Libraries. CoFreeUnused­Libraries would leave the COM shared data structures/components in a consistent state before unlocking the COM lock and then perform consistency checks after reacquiring the COM lock. COM architecture might precisely track each component's state to support its reentrant design. A state tracking mechanism could work like LDR_DDAG_STATE in the native loader. Co­Free­Unused­Libraries will acquire loader lock followed by the COM lock and then perform its task of freeing unused libraries. My inspiration for the second approach partially came from the GNU loader. In both solutions, the process maintains a single coherent lock hierarchy. Thus, either of these solutions would solve the potential for an ABBA deadlock upon calling CoFreeUnused­Libraries.

If some COM operation requires threads that load additional libraries (Is this something that can happen? As noted in our concurrency experiments, it's a naturally tricky situation that even the GNU loader will deadlock under. This is a tangent from the topic of ABBA deadlock.), then spawn that thread with the THREAD_CREATE_FLAGS_SKIP_LOADER_INIT flag (or remove thread blockers) and add library loading tasks to a global "load queue". The load queue will be similar to the existing work queue (which exists solely for parallelizing mapping and snapping work to improve performance) except for storing loader tasks for processing by the load owner (i.e. not just mapping or snapping). The current load owner (LoadOwner flag in TEB.SameTebFlags) will intermittently check the load queue and drain it on the other thread's behalf. The thread spawning a new thread would ideally check if it's under loader lock, and if so, wait on the new thread at least long enough to drain the new thread's load queue. If what libraries the new thread needs is known beforehand, then opt to load them on the load owner thread before spawning the new thread. Having a dedicated LoadOwner thread, provided suitable optimizations are in place, may also be a solution. DllMain can now safely run any COM code. Applying this solution to the CLR may mean removing .cctor because its continued presence means foreign (third-party) code could violate the loader lock -> CLR assembly initialization lock (I'm assuming the CLR loader has an assembly initialization lock similar to the native loader) hierarchy without our knowledge. The other solution is what Microsoft engineers currently advise, which is to write a rule saying that nobody should do COM (e.g. CoInitializeEx) from DllMain, which works great until people, such as Microsoft employees, do COM from DllMain (or one of many other places the loader unexpectedly holds loader lock) and Microsoft must reactively involve the CLR loader so "unmanaged and managed initialization is done in two separate and distinct stages" (introducing non-standard .cctor PE section) over the native loader to workaround "loader lock issues".

Given how monolithic Windows tends to be, the second solution for ABBA deadlock would be challenging; however, concurrency is hardly ever easy, and Microsoft engineers have taken on more complex problems. It's possible to make it work. The monolithic architecture of Windows is often what unexpectedly causes problems between these separate lock hierarchies to arise in the first place (e.g. due to a Windows API function internally doing COM without the programmer's explicit knowledge).

Note that the solutions I provide above are mostly thought experiments. Due to Windows' architectural issues, I can only offer bandage patches.

A Bonus Case Study on ABBA Deadlock Between the Loader Lock and GDI+ Lock

This case study is not regarding code that Microsoft engineers wrote but of a third-party shell extension that this Old New Thing article reports on. Here we have the ABBA deadlock between loader lock and GDI+ lock. The ABBA deadlock occurs more inadvertently than in the COM case study because of the GdiPlus!GdiplusShutdownthread waiting for a worker thread that holds the loader lock. Microsoft documentation and Raymond Chen say the root cause of this problem is runningGdiPlus!GdiplusStartupfromDllMain. However, my opinion on the root cause of this problem leans more on the architectural side of things, and it's that each library's DllMainreceivesDLL_THREAD_ATTACHandDLL_THREAD_DETACH` notifications in the first place. As noted elsewhere in this document, these notifications are effectively useless, and well-written libraries often disable them to improve performance. I'm also unaware of any other operating system requiring such notifications.

LdrpDrainWorkQueue and LdrpProcessWork

The high-level LdrpDrainWorkQueue and LdrpProcessWork parallel loader functions work hand-in-hand to perform module mapping and snapping.

The LdrpDrainWorkQueue function empties the work queue by processing all work in the work queue. LdrpDrainWorkQueue iterates over the ntdll!LdrpWorkQueue linked list, calling LdrpProcessWork on each work item to achieve its goal.

Processing a work item means doing mapping or snapping work on it, or both. LdrpProcessWork checks a module's LDR_DDAG_STATE. If it's zero (i.e. LdrModulesPlaceHolder), the module is unmapped, so LdrpProcessWork maps it; otherwise, LdrpProcessWork snaps it. In the mapping and snapping scenario, LdrpProcessWork calls LdrpMapDllFullPath or LdrpMapDllSearchPath. These mapping functions internally call through to the bloated (in my opinion, because it's doing things beyond what its name suggests) LdrpMapDllWithSectionHandle function. After mapping, LdrpMapDllWithSectionHandle decides whether also to do snapping by checking what LdrpFindLoadedDllByMappingLockHeld outputted for its fourth (out) argument (more research required). If this fourth argument outputs a non-zero value, LdrpMapDllWithSectionHandle calls LdrpLoadContextReplaceModule to enqueue the work (LdrpQueueWork) to the work queue; otherwise, LdrpMapDllWithSectionHandle adds the mapped module into the module information data structures (LdrpInsertDataTableEntry and LdrpInsertModuleToIndexLockHeld) before snapping. The loader can add a module's LDRP_LOAD_CONTEXT (undocumented structure; allocated by the LdrpAllocatePlaceHolder function) to the work queue before that module's LDR_DATA_TABLE_ENTRY has even been added to any of the module information data structures. The parallel loader worker threads may call LdrpWorkCallback -> LdrpProcessWork to pick up some mapping and/or snapping operations from the work queue as a performance optimization.

A work item refers to one LDRP_LOAD_CONTEXT entry (undocumented structure) in the loader work queue (ntdll!LdrpWorkQueue) linked list. Each work item relates directly to one module because they're allocated at the same time with the LdrpAllocatePlaceHolder function and each LDRP_LOAD_CONTEXT contains that a UNICODE_STRING of its own DLL's BaseDllName as its first member. However, processing a single work item for one module can still indirectly cause other modules to process within the same call to LdrpProcessWork because LdrpMapDllWithSectionHandle may call LdrpMapAndSnapDependency to process dependencies. Depending on the LDRP_LOAD_CONTEXT structure contents (undocumented), LdrpMapAndSnapDependency may outsource the work by enqueuing it (LdrpQueueWork) to the work queue (ntdll!LdrpWorkQueue) or process the dependency itself.

LdrpLoadCompleteEvent and LdrpWorkCompleteEvent

When the loader sets the LdrpLoadCompleteEvent event, it's signalling that the loader is reemerging from completing all mapping and/or snapping work for the entire work queue. When this occurs, it directly correlates with LdrpWorkInProgress equalling zero and the decommissioning of the current thread as the LoadOwner (see ntdll!LdrpDropLastInProgressCount function for evidence of this).

When the loader sets the LdrpWorkCompleteEvent event, it's signalling that the loader has completed the mapping and/or work of one work item.

Here are all the loader's usages of LdrpLoadCompleteEvent and LdrpWorkCompleteEvent:

0:000> # "ntdll!LdrpLoadCompleteEvent" <NTDLL ADDRESS> L9999999
ntdll!LdrpDropLastInProgressCount+0x38:
00007ffd`2896d9c4 488b0db5e91000  mov     rcx,qword ptr [ntdll!LdrpLoadCompleteEvent (00007ffd`28a7c380)]
ntdll!LdrpDrainWorkQueue+0x2d:
00007ffd`2896ea01 4c0f443577d91000 cmove   r14,qword ptr [ntdll!LdrpLoadCompleteEvent (00007ffd`28a7c380)]
ntdll!LdrpCreateLoaderEvents+0x12:
00007ffd`2898e182 488d0df7e10e00  lea     rcx,[ntdll!LdrpLoadCompleteEvent (00007ffd`28a7c380)]

0:000> # "ntdll!LdrpWorkCompleteEvent" <NTDLL ADDRESS> L9999999
ntdll!LdrpDrainWorkQueue+0x18:
00007ffd`2896e9ec 4c8b35bdd91000  mov     r14,qword ptr [ntdll!LdrpWorkCompleteEvent (00007ffd`28a7c3b0)]
ntdll!LdrpProcessWork+0x1e4:
00007ffd`2896ede0 488b0dc9d51000  mov     rcx,qword ptr [ntdll!LdrpWorkCompleteEvent (00007ffd`28a7c3b0)]
ntdll!LdrpCreateLoaderEvents+0x35:
00007ffd`2898e1a5 488d0d04e20e00  lea     rcx,[ntdll!LdrpWorkCompleteEvent (00007ffd`28a7c3b0)]
ntdll!LdrpProcessWork$fin$0+0x7c:
00007ffd`289b5ad7 488b0dd2680c00  mov     rcx,qword ptr [ntdll!LdrpWorkCompleteEvent (00007ffd`28a7c3b0)]

The LdrpCreateLoaderEvents function creates both events. The LdrpDrainWorkQueue function may wait on either LdrpLoadCompleteEvent or LdrpWorkCompleteEvent. Only the LdrpDrainWorkQueue function sets LdrpLoadCompleteEvent. Only the LdrpProcessWork function sets LdrpWorkCompleteEvent.

At event creation (NtCreateEvent), LdrpLoadCompleteEvent and LdrpWorkCompleteEvent are configured to be auto-reset events:

Auto-reset event objects automatically change from signaled to nonsignaled after a single waiting thread is released.

The LdrpDrainWorkQueue function calls NtWaitForSingleObject on these auto-reset events. The loader never manually resets the LdrpLoadCompleteEvent and LdrpWorkCompleteEvent events (NtResetEvent).

The LdrpDrainWorkQueue function takes one argument. This argument is a boolean, indicating whether the function should wait on LdrpLoadCompleteEvent or LdrpWorkCompleteEvent loader event before draining the work queue. Here's the relevant LdrpDrainWorkQueue IDA decompilation with marking up:

// Variable names are my own
struct _TEB *__fastcall LdrpDrainWorkQueue(BOOL IsProcessingSingleWorkItem)
{
    HANDLE EventHandle;
    BOOL RetryOrReturnEarly;
    CHAR LdrpDetourExistAtStart;
    PVOID LdrpWorkQueueNextEntry;
    ULONG_PTR LdrpWorkQueueLinkHolderCheckCorruptionTemp; // This is used for conducting a corruption check of the LdrpWorkQueue doubly linked list, terminating the offending process with RtlFailFast code FAST_FAIL_CORRUPT_LIST_ENTRY if so (used in the first piece of snipped code)
    BOOL LdrpDetourExistAtStartBoolean;
    PVOID CurrentTeb;
    ULONG_PTR RetryQueue; // The flink
    ULONG_PTR RetryQueueBlink; // Used in snipped code: "Add work item to LdrpRetryQueue"

    EventHandle = (HANDLE)LdrpWorkCompleteEvent;
    RetryOrReturnEarly = FALSE

    if ( !IsProcessingSingleWorkItem )
        eventHandle = LdrpLoadCompleteEvent;

    while ( TRUE )
    {
        while ( TRUE )
        {
            RtlEnterCriticalSection(&LdrpWorkQueueLock);
            // LdrpDetourExists relates to LdrpCriticalLoaderFunctions; see this document's section on that
            LdrpDetourExistAtStart = LdrpDetourExist;
            if ( !LdrpDetourExist || IsProcessingSingleWorkItem == TRUE )
            {
                // Notice that the LDRP_LOAD_CONTEXT entry flink is being dereferenced to get the next entry in the LdrpWorkQueue list
                LdrpWorkQueueNextEntry = (__int64 *)LdrpWorkQueue;
                // TODO: * Corruption check on LdrpWorkQueue list, modify list and other things*
                ... snip ...
            }
            else
            {
                if ( LdrpWorkInProgress == IsProcessingSingleWorkItem ) {
                    LdrpWorkInProgress = 1;
                    RetryOrReturnEarly = TRUE;
                }
                // NOTE: This path always sets LdrpWorkQueueNextEntry = LdrpWorkQueue so the following check on whether LdrpWorkQueue is empty will always pass
                LdrpWorkQueueNextEntry = &LdrpWorkQueue;
            }
            RtlLeaveCriticalSection(&LdrpWorkQueueLock);

            if ( RetryOrReturnEarly )
                break;
            // Tests if LdrpWorkQueue is empty
            // LdrpWorkQueueNextEntry points to the next LDRP_LOAD_CONTEXT entry in the LdrpWorkQueue list (at least if there is one)
            // If the LdrpWorkQueue list head equals the next entry in the circular linked list, the list must be empty
            if ( &LdrpWorkQueue == LdrpWorkQueueNextEntry )
            {
                NtWaitForSingleObject(EventHandle, 0, NULL);
            }
            else
            {
                // Cast to a boolean (LdrpDetourExist is always a boolean value anyway, but explictly casting it as the boolean type is required for the function parameter)
                // Note that this actually casts to the lowest one byte because you can't just pass a single bit to a function
                LOBYTE(LdrpDetourExistAtStartBoolean) = LdrpDetourExistAtStart;
                // LdrpProcessWork processes (i.e. mapping and snapping) the specified work item
                LdrpProcessWork(LdrpWorkQueueNextEntry - 8, LdrpDetourExistAtStartBoolean);
            }
        }
  
        // Check whether to retry using ntdll!LdrpRetryQueue list (I haven't looked into this)
        // Test if not IsProcessingSingleWorkItem OR LdrpRetryQueue is empty
        if ( !IsProcessingSingleWorkItem || &LdrpRetryQueue == (__int64 *)LdrpRetryQueue )
            break;
        RtlEnterCriticalSection(&LdrpWorkQueueLock);
        RetryQueue = LdrpRetryQueue;
        // TODO: * Add work item to LdrpRetryQueue code *
        ... snip ...
        RtlLeaveCriticalSection(&LdrpWorkQueueLock);
        RetryOrReturnEarly = FALSE;
    }

    // Set current thread as the load owner so the rest of the loading process can take place upon returning
    // This happens *after* processing (i.e. mapping and snapping) work
    // Note that parallel loaders threads never call LdrpDrainWorkQueue (they do LdrpWorkCallback -> LdrpProcessWork)
    // Likely retuns a pointer to the TEB because that's faster than having the caller retrieve the TEB again
    CurrentTeb = NtCurrentTeb();
    CurrentTeb->SameTebFlags |= 0x1000u;
    return CurrentTeb;
}

Notice the infinite loop immediately around waiting for an event. This loop is necessary because the loader never manually resets an event, which means an event could be left over as set from a previous loader operation. Thus, NtWaitForSingleObject would allow execution to proceed to the next loop around, and if LdrpDrainWorkQueue doesn't break out of the loop (break) this time, LdrpDrainWorkQueue will wait on the now waiting loader event. I've seen this very thing occur during LdrpDrainWorkQueue run-time. Therefore, how LdrpDrainWorkQueue synchronizes on a loader event may cause it to operate effectively like a spinlock for one iteration.

What follows documents what parts of the loader call LdrpDrainWorkQueue while synchronizing on either the LdrpLoadCompleteEvent or LdrpWorkCompleteEvent loader event (data gathered by searching disassembly for calls to the LdrpDrainWorkQueue function):

ntdll!LdrUnloadDll+0x80: FALSE
ntdll!RtlQueryInformationActivationContext+0x43c: FALSE
ntdll!LdrShutdownThread+0x98: FALSE
ntdll!LdrpInitializeThread+0x86: FALSE
ntdll!LdrpLoadDllInternal+0xbe: FALSE
ntdll!LdrpLoadDllInternal+0x144: TRUE
ntdll!LdrpLoadDllInternal$fin$0+0x38: TRUE
ntdll!LdrGetProcedureAddressForCaller+0x270: FALSE
ntdll!LdrEnumerateLoadedModules+0xa7: FALSE
ntdll!RtlExitUserProcess+0x23: TRUE OR FALSE
  - Depends on `TEB.SameTebFlags`, typically FALSE if `LoadOwner` or `LoadWorker` flags are absent, TRUE if either of these flags is present
ntdll!RtlPrepareForProcessCloning+0x23: FALSE
ntdll!LdrpFindLoadedDll+0x9127a: FALSE
ntdll!LdrpFastpthReloadedDll+0x9033a: FALSE
ntdll!LdrpInitializeImportRedirection+0x46d44: FALSE
ntdll!LdrInitShimEngineDynamic+0x3c: FALSE
ntdll!LdrpInitializeProcess+0x130a: FALSE
ntdll!LdrpInitializeProcess+0x1d0d: FALSE
ntdll!LdrpInitializeProcess+0x1e22: TRUE
ntdll!LdrpInitializeProcess+0x1f33: FALSE
ntdll!RtlCloneUserProcess+0x71: FALSE

Notably, there are many more instances of synchronizing on the completion of processing all work in the work queue rather than just one work item. For example, thread initialization (LdrpInitializeThread) synchronizes on the former (LdrpLoadCompleteEvent). The only functions that may synchronize on a single work item are LdrpLoadDllInternal (the public LoadLibrary function internally calls this), LdrpInitializeProcess, and RtlExitUserProcess.

Comparing the loader's number of calls to LdrpDrainWorkQueue in contrast to directly calling LdrpProcessWork, we see the former is considerably more prevalent:

0:000> # "call    ntdll!LdrpProcessWork" <NTDLL ADDRESS> L9999999
ntdll!LdrpLoadDependentModule+0x184c:
00007ffb`1ca8942c e8cb570400      call    ntdll!LdrpProcessWork (00007ffb`1cacebfc)
ntdll!LdrpLoadDllInternal+0x13a:
00007ffb`1ca8fb4e e8a9f00300      call    ntdll!LdrpProcessWork (00007ffb`1cacebfc)
ntdll!LdrpDrainWorkQueue+0x17f:
00007ffb`1caceb53 e8a4000000      call    ntdll!LdrpProcessWork (00007ffb`1cacebfc)
ntdll!LdrpWorkCallback+0x6e:
00007ffb`1cacebde e819000000      call    ntdll!LdrpProcessWork (00007ffb`1cacebfc)

There are only three occurrences of the loader directly calling LdrpProcessWork (not including the call from LdrpDrainWorkQueue) versus the twenty times it calls LdrpDrainWorkQueue.

When the Process Would Rather Terminate Then Wait on a Critical Section

In the documentation for EnterCriticalSection is this interesting snippet:

While a process is exiting, if a call to EnterCriticalSection would block, it will instead terminate the process immediately. This may cause global destructors to not be called.

Curious to see how Microsoft implemented this, I found that RtlpWaitOnCriticalSection checks if PEB_LDR_DATA.ShutdownInProgress is true. If so, the function jumps to calling NtTerminateProcess passing -1 as the first argument, thereby forcefully terminating the process.

PEB_LDR_DATA.ShutdownInProgress impacts several things, including FLS callbacks at process exit, TLS destructors at process exit, and DLL_PROCESS_DETACH notifications to DllMain at process exit.

The LdrShutdownProcess function (called by RtlExitUserProcess) sets PEB_LDR_DATA.ShutdownInProgress to true.

ABBA Deadlock

An ABBA deadlock is a deadlock due to lock hierarchy violation; whether a lock hierarchy violation (i.e. lock ordering inversion) results in an ABBA deadlock is probabilistic.

A more complex ABBA deadlock can manifest when nesting separate lock hierarchies within a thread. Suppose another thread interacts with the same lock hierarchies in the opposite order. In that case, the threads acquiring these locks can interleave (i.e. lock ordering inversion between lock hierarchies), thus leading to deadlock. This case can be tricky because there may not necessarily be a defined order for correctly acquiring the locks between separate lock hierarchies.

Although the ABBA deadlock and the ABA problem names may sound similar, they have entirely different meanings, so be sure to distinguish between them.

ABA Problem

The ABA problem ABA problem is a lower-level concept that can cause data structure inconsistency in lock-free code (i.e. code that relies entirely on atomic assembly instructions to ensure data structure consistency):

In multithreaded computing, the ABA problem occurs during synchronization, when a location is read twice, has the same value for both reads, and the read value being the same twice is used to conclude that nothing has happened in the interim; however, another thread can execute between the two reads and change the value, do other work, then change the value back, thus fooling the first thread into thinking nothing has changed even though the second thread did work that violates that assumption.

A common case of the ABA problem is encountered when implementing a lock-free data structure. If an item is removed from the list, deleted, and then a new item is allocated and added to the list, it is common for the allocated object to be at the same location as the deleted object due to MRU memory allocation. A pointer to the new item is thus often equal to a pointer to the old item, causing an ABA problem.

This example shows how atomic compare-and-swap atomic instructions (lock cmpxchg in x86) could mix up list items on concurrent deletion and creation because a new list item allocation at the same place in memory as the just deleted list item (e.g., in a linked list) could interleave, thus causing the ABA problem.

As a result, the risk of naively programmed lock-free code realizing the ABA problem is probabilistic. It's highly probabilistic with dynamically allocated memory, particularly when considering that modern heap allocators avoid returning the same block of memory in too close succession as a form of exploit mitigation (i.e. modern heap allocators may not do MRU memory allocation as the ABA problem Wikipedia page suggests).

Modern instruction set architectures (ISAs), such as AArch64, support load-link/store-conditional (LL/SC) instructions, which provide a stronger synchronization primitive than compare-and-swap (CAS). LL/SC solves the ABA problem on supporting architectures.

Although the ABA problem and the ABBA deadlock names may sound similar, they have entirely different meanings, so be sure to distinguish between them.

Dining Philosophers Problem

The dining philosophers' problem is a classic scenario originally devised by Dijkstra to illustrate synchronization issues that occur in concurrent algorithms and how to resolve them. Here's the problem statement.

The simplest solution is that philosophers pick a side (e.g., the left side) and then agree always to take a fork from that side first. If a philosopher picks up the left fork and then fails to pick up the right fork because someone else is holding it, he puts the left fork back down. Otherwise, now holding both forks, he eats some spaghetti and then puts both forks back down. By controlling the order in which philosophers pick up forks, you effectively create lock hierarchies between the left and right forks, thus preventing deadlock.

Anyone who's learned about concurrency throughout their computer science program would be familiar with this problem.

Compiling & Running

1. Run make where there is a Makefile
2. Run source ../set-ld-lib-path.sh
3. Run ./main or gdb ./main to debug

Analysis Commands

These are some helpful tips and commands to aid analysis.

LDR_DATA_TABLE_ENTRY Analysis

List all module LDR_DATA_TABLE_ENTRY structures:

!list -x "dt ntdll!_LDR_DATA_TABLE_ENTRY" @@C++(&@$peb->Ldr->InLoadOrderModuleList)

LDR_DDAG_NODE Analysis

List all module DdagNode structures:

!list -x "r $t0 =" -a "; dt ntdll!_LDR_DATA_TABLE_ENTRY @$t0 -t BaseDllName; dt ntdll!_LDR_DDAG_NODE @@C++(((ntdll!_LDR_DATA_TABLE_ENTRY *)@$t0)->DdagNode)" @@C++(&@$peb->Ldr->InLoadOrderModuleList)

List all module DdagNode.State values:

!list -x "r $t0 =" -a "; dt ntdll!_LDR_DATA_TABLE_ENTRY @$t0 -cio -t BaseDllName; dt ntdll!_LDR_DDAG_NODE @@C++(((ntdll!_LDR_DATA_TABLE_ENTRY *)@$t0)->DdagNode) -cio -t State" @@C++(&@$peb->Ldr->InLoadOrderModuleList)

Trace all DdagNode.State values starting from its initial allocation to the next module load for every module:

bp ntdll!LdrpAllocateModuleEntry "bp /1 @$ra \"bc 999; ba999 w8 @@C++(&((ntdll!_LDR_DATA_TABLE_ENTRY *)@$retreg)->DdagNode->State) \\\"ub . L1; g\\\"; g\"; g"

• This command breaks on ntdll!LdrpAllocateModuleEntry, sets a temporary breakpoint on its return address (@$ra), then continuing execution, sets a watchpoint on DdagNode.State in the returned LDR_DATA_TABLE_ENTRY, and log the previous disassembly line on watchpoint hit
• We must clear the previous watchpoint (bc 999) to set a new one every time a new module load occurs (ModLoad message in WinDbg debug output). Deleting hardware breakpoints is necessary because they are a finite resource.

Throw this in with the trace command after ub . L1; to check some locks: !critsec LdrpLoaderLock; dp LdrpModuleDataTableLock L1;

Warning: Due to how module loads work out, some state changes will never appear in a trace due to deleting watchpoints. For example, this trace command won't uncover a LdrModulesSnapping state change in the LdrpSignalModuleMapped function or a LdrModulesWaitingForDependencies state change. For this reason, it's necessary to sample modules randomly over a few separate executions. Run sxe ld:<DLL_TO_BEGIN_SAMPLE_AT> and remove the watchpoint deletion command (bc 999), then run the trace command starting from that point.

Warning: This command will continue tracing an address after a module unloads, which causes LdrpUnloadNode -> RtlFreeHeap of memory. Disregard junk results from that point forward due to potential reallocation of the underlying memory until the next ModLoad WinDbg message, meaning a fresh watchpoint.

Trace TEB.SameTebFlags

ba w8 @@C++(&@$teb->SameTebFlags-3)

TEB.SameTebFlags isn't memory-aligned, so we need to subtract 3 to set a hardware breakpoint. This watchpoint still captures the full TEB.SameTebFlags but now TEB.MuiImpersonation, TEB.CrossTebFlags, and TEB.SpareCrossTebBits, in front of TEB.SameTebFlags in the TEB structure, will also fire our watchpoint. However, these other members are rarely used, so it's only a minor inconvenience.

ASLR randomizes the TEB's location on every execution, so you must set the watchpoint again when restarting the program in WinDbg.

Read the TEB: dt @$teb ntdll!_TEB

Searching Assembly for Structure Offsets

These commands contain the offset/register values specific to a 64-bit process. Please adjust them if you're working with a 32-bit process.

WinDbg command: # "*\\+38h\\]*" <NTDLL ADDRESS> L9999999

• Search NTDLL disassembly for occurrences of offset +0x38 (useful to search for potential LDR_DDAG_NODE.State references)
• Put the output of this command into an offset-search.txt file

Filter command pipeline (POSIX sh): sed '/^ntdll!Ldr/!d;N;' offset-search.txt | sed 'N;/\[rsp+/d;'

• The WinDbg # command outputs two lines for each finding, hence the sed commands using its N command to read the next line into pattern space
• First command: Filter for findings beginning with ntdll!Ldr because we're interested in the loader
  • Use this sed as a grep -v -A1 substitute because the latter prints a group separator, which we don't want (GNU grep supports --no-group-separator to remove this, but I prefer to keep POSIX compliant)
• Second command: Filter out offsets to the stack pointer, which is just noise operating on local variables

Depending on what you're looking for, you should filter further—for example, filtering down to only mov or lea assembly instructions. Be aware that the results won't be comprehensive because the code could also add/subtract an offset from a member in the structure it's already at to reach another member in the current structure. It's still generally helpful, though.

Due to two's complement, negative numbers will, of course, show up like: 0FFFFFFFFh (e.g. for -1, WinDbg disassembly prints it with a leading zero)

Monitor a Critical Section Lock

Watch for reads/writes to a critical section's locking status.

ba r4 @@C++(&((ntdll!_RTL_CRITICAL_SECTION *)@@(ntdll!LdrpDllNotificationLock))->LockCount)

I tested placing a watchpoint on loader lock (LdrpLoaderLock). Doing this on a modern Windows loader won't tell you much because, as previously mentioned, it's the state, such as the LDR_DDAG_NODE.State or LoadOwner in TEB.SameTebFlags that the loader internally tests to make decisions. Setting this watchpoint myself during complex operations, I've never seen an occurrence of Windows using the ntdll!RtlIsCriticalSectionLocked or ntdll!RtlIsCriticalSectionLockedByThread functions to branch on the state of loader lock itself.

Track Loader Events

This WinDbg command tracks load and work completions. It helped figure out the difference between these two loader events.

ba e1 ntdll!NtSetEvent ".if ($argreg == poi(ntdll!LdrpLoadCompleteEvent)) { .echo Set LoadComplete Event; k }; .elsif ($argreg == poi(ntdll!LdrpWorkCompleteEvent)) { .echo Set WorkComplete Event; k }; gc"

We use a hardware execution breakpoint (ba e1) instead of a software breakpoint (bp) because otherwise, there's some strange bug where WinDbg may never hit the breakpoint for LdrpWorkCompleteEvent (root cause requires investigation).

Note that this tracer is slow because Windows often uses events (calling NtSetEvent) even outside the loader.

Swap out ntdll!NtSetEvent for ntdll!NtWaitForSingleObject and change echo messages to get more info. The LdrpLoadCompleteEvent and LdrpWorkCompleteEvent events are never manually reset (ntdll!NtResetEvent). They're auto-reset events, so passing them into a wait function causes them to reset (even if no actual waiting occurs).

Be aware that anti-malware software hooking on ntdll!NtWaitForSingleObject could cause the breakpoint command to run twice (I ran into this issue).

View state of Win32 events WinDbg command: !handle 0 8 Event

Disable Loader Worker Threads

This CMD command helps analyze what locks are acquired in what parts of the code without the chance of a loader thread muddying your analysis (among other things). These show up as ntdll!TppWorkerThread threads in WinDbg at process startup (note that any threads belonging to other thread pools also take this thread name).

reg add "HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Image File Execution Options\<YOUR_EXE_FILENAME>" /v MaxLoaderThreads /t REG_DWORD /d 1 /f

Yes, the correct value for disabling loader threads from spawning is 1. Set 0 and loader threads still spawn (0 appears to be the same as unset). Setting 2 spawns one loader work thread, and that pattern continues...

Internally, at process initialization (LdrpInitializeProcess), the loader retrieves MaxLoaderThreads in LdrpInitializeExecutionOptions. LdrpInitializeExecutionOptions queries this registry value (along with others) and saves the result. Later, the MaxLoaderThreads registry value becomes the first argument to the LdrpEnableParallelLoading function. LdrpEnableParallelLoading validates the value before passing it as the second argument to the TpSetPoolMaxThreads function. The TpSetPoolMaxThreads function does a NtSetInformationWorkerFactory system call with the second argument being the enum value WorkerFactoryThreadMaximum (5) to set the maximum thread count and the desired count pointed to by the third argument within the WorkerFactoryInformation structure (undocumented).

The maximum number of threads the parallel loader allows is 4. Specifying a higher thread count will top out at 4.

Creating threads happens with ntdll!NtCreateWorkerFactory. The ntdll!NtCreateWorkerFactory function only does the NtCreateWorkerFactory system call; this is a kernel wrapper for creating multiple threads at once, improving performance because it avoids extra user-mode <-> kernel-mode context switches.

Some operations, such as ShellExecute, may spawn threads named ntdll!TppWorkerThread. However, these worker threads belong to an unrelated thread pool, not the loader work thread pool, and thus, never call the ntdll!LdrpWorkCallback function entry point. Make sure to distinguish loader work thread pools: dt @$teb ntdll!_TEB -t LoaderWorker

List All LdrpCriticalLoaderFunctions

Whether a critical loader function (ntdll!LdrpCriticalLoaderFunctions) is detoured (i.e. hooked) determines whether ntdll!LdrpDetourExist is TRUE or FALSE. BlackBerry's Jeffrey Tang covered this in "Windows 10 Parallel Loading Breakdown"; however, this article only lists the first few critical loader functions, so here we show all of them. Below are the commands for determining all critical loader functions should they change in the future:

0:000> ln ntdll!LdrpCriticalLoaderFunctions
(00007ffb`1cb8df20)   ntdll!LdrpCriticalLoaderFunctions   |  (00007ffb`1cb8e020)   ntdll!RtlpMemoryZoneCriticalRoutines
0:000> ? (00007ffb`1cb8e020 - 00007ffb`1cb8df20) / @@(sizeof(void*))
Evaluate expression: 32 = 00000000`00000020
0:000> .for (r $t0=0; $t0 < <OUTPUT_OF_PREVIOUS_EXPRESSION_HERE>; r $t0=$t0+1) { u poi(ntdll!LdrpCriticalLoaderFunctions + $t0 * @@(sizeof(void*))) L1 }
ntdll!NtOpenFile:
00007ffb`1cb0d630 4c8bd1          mov     r10,rcx
ntdll!NtCreateSection:
00007ffb`1cb0d910 4c8bd1          mov     r10,rcx
ntdll!NtQueryAttributesFile:
00007ffb`1cb0d770 4c8bd1          mov     r10,rcx
... more output ...

Clean up the output by filtering it through this POSIX sed pipeline: sed 'n;d;' <OUTPUT_FILE> | sed 's/.$//'

Finally, we have a comprehensive list of critical loader functions:

ntdll!NtOpenFile
ntdll!NtCreateSection
ntdll!NtQueryAttributesFile
ntdll!NtOpenSection
ntdll!NtMapViewOfSection
ntdll!NtWriteVirtualMemory
ntdll!NtResumeThread
ntdll!NtOpenSemaphore
ntdll!NtOpenMutant
ntdll!NtOpenEvent
ntdll!NtCreateUserProcess
ntdll!NtCreateSemaphore
ntdll!NtCreateMutant
ntdll!NtCreateEvent
ntdll!NtSetDriverEntryOrder
ntdll!NtQueryDriverEntryOrder
ntdll!NtAccessCheck
ntdll!NtCreateThread
ntdll!NtCreateThreadEx
ntdll!NtCreateProcess
ntdll!NtCreateProcessEx
ntdll!NtCreateJobObject
ntdll!NtCreateEnclave
ntdll!NtOpenThread
ntdll!NtOpenProcess
ntdll!NtOpenJobObject
ntdll!NtSetInformationThread
ntdll!NtSetInformationProcess
ntdll!NtDeleteFile
ntdll!NtCreateFile
ntdll!NtMapViewOfSectionEx
ntdll!NtExtendSection

Make sure to prepend 0n to OUTPUT_OF_PREVIOUS_EXPRESSION_HERE if you use the base ten representation (0n32); otherwise, WinDbg interprets that number as hex (base 16) by default.

Microsoft's public symbols only come with names that typically do not include any information on type or size. Hence, we must find the ntdll!LdrpCriticalLoaderFunctions size with the ln command instead of sizeof.

Get TCB and Set GSCOPE watchpoint (GNU Loader)

Read the thread control block on GNU/Linux (equivalent to TEB on Windows):

set print pretty
print *(tcbhead_t*)($fs_base)

Set a watchpoint on this thread's GSCOPE flag:

print &((tcbhead_t*)($fs_base))->gscope_flag
watch *<OUTPUT_OF_LAST_COMMAND>

Debugging Notes

Ensure you have glibc debug symbols and, preferably, source code downloaded. Fedora's GDB automatically downloads symbols and source code. On Debian, you must install the libc6-dbg (you may need to enable the debug source in your /etc/apt/sources.list) and glibc-source packages.

When printing a backtrace, you may see <optimized out> in the output for some function parameter values. Seeing these requires compiling a debug build of glibc. Not debug symbols, a debug build. You could also set a breakpoint where the value is optimized out to see it. Generally, you don't need to see these values, though, so you can ignore them.

In backtraces contained within the gdb-log.txt files of this repo, you may entries to functions like dlerror_run, _dl_catch_error and __GI__dl_catch_exception (_dl_catch_exception in the code). These aren't indicative of an error occurring. Rather, these functions merely set up an error handler and perform handling to catch exceptions if an error/exception occurs in the dynamic linker. The error codes are conformant to errno. The most common error code is an ENOMEM (out of memory), but a file access error code (ENOENT or EACCES) can also occur. Dynamic linker functions may also pass 0 as the errno to indicate an error specific to the dynamic linker's functionality. In either case, an exception occurs, thus stopping the dynamic linker from proceeding and continuing execution in _dl_catch_exception (called by _dl_catch_error). A programmer can determine if dlopen failed by checking for a NULL return value and then get an error message with dlerror.

Big thanks to Microsoft for successfully nerd sniping me!

The document you just read is under a CC BY-SA License.

This repo's code is under the MIT License.

Copyright (C) 2023-2024 Elliot Killick [e-mail]

EOF