My program depends on a 3rd party static library. The library will start a thread triggered by the initialization of a global variable and the thread will lock on a mutex, and I can't change the 3rd party's code. In my program inside main(), I will invoke fork() and exit to make my program a daemon.
The behavior described above may result in dead lock because the lock can't get unlocked in the child process forked when the global variable is being uninitialized after main().
So now what I want to do is to fork the process before the 3rd party static library is loaded. So is there a proper way to achieve this?
So now what I want to do is to fork the process before the 3rd party static library is loaded. So is there a proper way to achieve this?
"Before the static library is loaded" is not a thing, static libraries are linked in at build time, not loaded dynamically at runtime.
The library will start a thread triggered by the initialization of a global variable
You should consider using a better-designed library, or work with the maintainer to come up with a better-thought-out initialization scheme. Life before main is usually a bad idea. Spawning threads before main is a really bad idea. Alternatives might be, use a lazy static that gets triggered whenever an API of the library is called, or better yet, introducing an explicit init function, or using context pointers so that the user can manage library state. Sorry, it sounds like this thing is a tire fire and theres no sane way to fork it until you invest some time in fixing it.
Use another program that does the daemonization and then execs your program.
Some service supervisors specifically require that their services do not double fork. So that they can detect when the service exits. As being ran under the supervisor already means that they don't have a controlling terminal so there is no need to detach.
Related
I want to unload a DLL from another process. Is it possible?
If yes, how to do that? (I'm using Win32 API)
Thank you very much.
Yes, this is possible. It is called DLL ejection and is featured by some DLL injectors. The way a DLL is usually loaded is via LoadLibrary and it is subsequently unloaded via FreeLibrary. FreeLibrary takes only one parameter which is a handle to the module to be unloaded. If you injected the DLL in the first place, you should be able to find this very easily. Otherwise there are ways of obtaining the handle such as CreateToolHelp32Snapshot with further enumeration with Module32First/Module32Next. Suppose you have obtained the handle through some means, then the steps to eject the DLL are simple:
Get the address of FreeLibrary with GetProcAddress. This address will match the one for the same function in the target because of how Windows works.
Call CreateRemoteThread on the target process, specifying lpStartAddress as the address of FreeLibrary, and with lpParameter as the module's handle
There are several caveats to DLL ejection.
You should only ever eject a DLL which you are certain no code is going to make use of again in the future. If any dynamically linked code attempts to make a call to your code after it has been freed, it will most likely trigger some form of page access violation.
You should ensure that no threads are executing within the code of the DLL whilst ejection is being performed for similar reasons.
DLL ejection should be avoided with general. If the library wants to have the option of being freed, it should supply some interface which users can access it through which eventually calls FreeLibraryAndExitThread.
If you require a code example for this, I have written an ejector as part of an injector I wrote in the past in C. I can search it up and find it but it's from many years ago and the code quality is not likely to be good.
You don't want to do this.
"Loading" a DLL is much more than simply opening (and locking) a file. When the NT loader starts an executable, it processes all the DLLs referenced by the image (recursively) and wires up the function calls (recursively): loading the DLLs, calling the DLL initialization code, etc.
Unloading a DLL would mean that you'd need to stop all processes that loaded your DLL, load the new DLL, and perform all the operations the NT loader would. Of course, unloading and re-loading a DLL would need to restore that old DLL's state (initialized variables etc), an operation which is not specified in Win32.
For a bit of background information, see this article on MSDN and this Under the Hood article in MSJ.
Short answer: No, it is impossible.
Win32 doesn't provide an API to unload a DLL of another process. If a library is freed unexpectedly, the process will crash. This leads to a serious security hole as it breaks process protection mechanism.
If you can modify both of the processes, you can modify the application and add routines to free a library, and let the other application to send the message.
I would instead look to change the function called when the process tries to invoke the functions in that dll. I know this is possible in theory.
It would mean a bit of memory hacking and knowing where the pointers to the functions are stored, but all of that can be found easily enough (ollydbg manages to do it), it would be harder if they use ordinals, even harder if they hard code the pointers, but no one does that nowadays. You could then inject your own code that (ideally) mimics the functions they mask, but does not actually do anything. They will probably have to be injected into the process, and that way you could get it to work without the process ever knowing, and without any crashes.
Here is a little background information. I'm working on replacing a dll that has been used in a dll injection technique via the AppInit_DLLs registry entry. Its purpose was to be present in every process and set hooks into the GDI32.dll to gather information about printing. This is kind of a funky way to get what we want. The .dll itself is over 10 years old (written in Visual Studio 97) and we'd like to replace it with something a little less invasive than an injected dll.
It appears SetWindowsHookEx() maybe what we are looking for. I've been having some trouble with it, but I've also had some discussions with co-workers about whether or not this tree is worth barking up. Here are some questions that we could not determine:
When we hook a routine out of a dll, for example StartDoc() from GDI32.dll, do we really get a notification every time any other process uses that rotuine out of that dll? This is kind of the functionality we were getting with our injected .dll and we need the same functionality going forward.
When the hook is triggered, does the hook handling procedure run in the process space of the process that initiated the actual call, or in the process space of the process that set up the hook? My opinion is that it has to run in the process space of the process that called the routine. For example, if a program calls StartDoc() from GDI32.dll, it will have the hook handling procedure code "injected" into its space and executed. Otherwise, there would have to be some inter-process communication that automatically gets set up between the calling process and the process that set up the hook, and I just don't see that as being the case. Also, its kind of necessary that this hook handling routine run in the process space of the calling process since one of the things it needs to know is the name of that calling process, and I'm not sure on how to get that information if it wasn't actually running in that process.
If the hook handling routine is written using the .NET managed environment, will it break when getting hooked into a process not utilizing the .NET managed environment? We'd really like to move away from C++ here and use C#, but what would happen if we our hook gets called from a process that isn't managed? As stated before, I think that our hook handling procedure will run in the process that originally called the routine that was hooked. But if this is true, than I would think that we'd run into trouble if this process was not using the .NET run time environment, but the incoming hooked handling code is.
Yes.
Generally, it's the former: it executes in the context of the process whose event it is hooking.
After a successful call to SetWindowsHookEx, the operating system automatically injects the hook DLL (the one that contains the callback function) into the address space of all target processes that meet the requirements for the specified hook type. (Of course, the hooking code is not necessarily injected immediately.)
The exception to this general rule are the low-level keyboard and mouse hooks (WH_LL_KEYBOARD and WH_LL_MOUSE). Since those hook types are not injected into the client processes, the callback is called in the same thread that originally called SetWindowsHookEx.
That last point is important to keep in mind to answer your third question. Because the low-level keyboard and mouse hooks are the only two global hooks that do not require DLL injection, they are also the only two types of hooks that can be written in managed .NET code.
For the other hook types, your concerns expressed in the question are precisely correct. You would need to write these hook DLLs in C or C++. Of course, the rest of your application's pieces could still be written in a managed language. The only thing that matters is the hook DLL.
You might consider looking into either Microsoft Detours or EasyHook.
Recently I have been reading up articles about DLL injection and I understand them fairly well.
However, what I don't understand is why APIs such as CreateRemoteThread, WriteProcessMemory(in being able to write to the memory of another process) and VirtualAllocEx(in being able to allocat memory in the context of another process) were implemented in the first place.
What was the original need for such APIs? Just curious.
WriteProcessMemory was made for ring3 debuggers that need to securely write process memory, most commonly for INT 3 breakpoints or user provided memory edits.
along the same line, CreateRemoteThread can also be used for debugging purposes, however, MSDN can enlighten us on CreateRemoteThread a bit more:
A common use of this function is to inject a thread into a process
that is being debugged to issue a break. However, this use is not
recommended, because the extra thread is confusing to the person
debugging the application and there are several side effects to using
this technique:
It converts single-threaded applications into
multithreaded applications.
It changes the timing and memory layout of
the process.
It results in a call to the entry point of each DLL in
the process.
IIRC, CreateRemoteThread is also used by debuggers to hook application native expection handlers, commonly set by SetExceptionHandler, which requires call from the target process as the handler is stored in the PEB.
VirtualAllocEx is just how windows virtual memory system operates, it needs a context to allocate in, be it in the current process, a child process or a remote process. VirtualAlloc in fact is nothing more than a pass through wrapper of the Ex variant, it just passes a special constant that indicates the handle of the caller process is to be used.
In my Cocoa application I need to run a task that uses unstable unfixable code. It takes little input, works independently from the rest of the app, and generates thousands of ObjC objects as a result.
How can I run the unstable part and let it crash without taking down whole application?
Is it possible to fork() Cocoa application? How UI, threads, GC, KVO, runloops are going to behave when forked?
Can I avoid creating standalone executable launched via NSTask?
If I launch separate process, how can I send and receive ObjC object instances? (I'd rather not serialize/unserialize them myself, and I need to keep them after child process ends).
How does OS X handle this problem for Spotlight and Quicklook plugins?
Is it possible to fork() Cocoa application?
Yes, but you pretty much have to exec immediately. Core Foundation will throw an exception if you try to use certain Cocoa methods or CF functions between fork and exec (or without execking at all). You might get away with some things (I was able to ask a window its frame, for example), but nothing is safe.
Launching an NSTask, of course, counts as fork and exec together, averting the problems of skipping or postponing the exec.
How UI, threads, GC, KVO, runloops are going to behave when forked?
UI: Windows (the actual ones on the screen) are not duplicated. Of course, you can't talk to your NSWindow and NSView objects anyway.
Threads: Not carried over to the subprocess. This is not as good as it may sound, as problem cases abound; for one, another thread might have held a lock in the parent, which remains locked in the child even though the thread that held it is absent.
GC: Well, the garbage collector runs on a thread…
KVO: Should be fine, since observation is normally triggered either explicitly or by KVO-supplied wrapper accessors.
Run loops: One per thread, so the main thread's run loop should still exist, but it will die if you return to it.
Can I avoid creating standalone executable launched via NSTask?
Nope.
If I launch separate process, how can I send and receive ObjC object instances?
If you don't exec, you don't.
Otherwise, you can use DO.
(I'd rather not serialize/unserialize them myself, and I need to keep them after child process ends).
Then you'll need to make a copy in the parent process. I don't know whether you can use copyWithZone: here; probably not. I suspect you will have to do some sort of plist- or archive-based serialization/unserialization.
How does OS X handle this problem for Spotlight and Quicklook plugins?
Spotlight has mdworker; Quick Look has something similar.
I use Distributed Objects to communicate between my cocoa program and a separate (unreliable) worker program. I start the worker as a NSTask. Distributed objects are very elegantly put together.
My Win32 console applicaton uses a third-party library. After it exits WinMain global objects destruction begins and an AV happens somewhere deep inside. I'm really tempted to just write
TerminateProcess( GetCurrentProcess(), 0 );
somewhere near the end of WinMain. If I do this the application ends gracefully.
But MSDN says that doing so can compromise the state of global data maintained by dynamic-link libraries (DLLs) which is not clear. I understand that if I have some global object its destructor is not run and I risk not finalizing a database connection or something similar. I don't have anything like that in my program.
What exactly is the risk when using TerminateProcess? How do I determine if I can use it for my purpose?
Based on the documentation for that and ExtiProcess it seems the primary concern is that DLL's are unloaded without a call to DllMain with the flag DLL_PROCESS_DETACH.
My 2cents: The documentation is being paranoid that you will upset some critical operation which runs in DllMain + DLL_PROCESS_DETACH. Anyone who depends on that to maintain critical state is already at the mercy of task manager so I don't see a huge risk in using this API.
Generally the bad things will happen when interacting with objects outside of your process. For an example say you have some shared memory used by multiple processes that your process will write to and other processes read and or write to. Typically to synchronize the reading and writing a mutex is used. If a thread in your process has acquired the mutex and is in the middle of making changes when TerminatePorcess is called, the mutex will be abandoned and the shared memory potentially left in an inconsistent state.
I suspect you are miss using one of the third party libraries. DllMain is somewhat limiting so the library may have initialize and uninitialize functions that you are supposed to call.
AFAIK, if you're not doing anything "fancy" (which includes but is not limited to: creating threads, locks, DB connections, using COM objects), nothing terrible will happen. But as Earwicker says, you don't know what OS-wide stuff a DLL is doing, and you certainly don't know if that will change in the future, so relying on this is very fragile.
Aren't you curious to know why this access violation is occurring? It may well be the sign of something that became corrupted much earlier on. Please at least confirm that the bug is caused by this 3rd-party library, e.g. by writing a program that links with the library but whose main() does nothing, and confirming that this causes the same crash.
It depends how you interpret "global data". If you take it to mean (as I normally would) data stored in the process's address space, then the advice makes no sense - we know that memory is going to disappear, so who cares what happens to that?
So it may be referring to OS-wide stuff that a DLL may have done, that persists outside the lifetime of any process. A simple example would be a temporary file that might need to be cleaned up; crash the process too many times and you'll run out of disk space, so probably best not to make a habit of it.