I was always wondering, when I press a key on my keyboard, windows sees it as KEY_DOWN event, and then my program can ask windows if there is such an event, and the windows will answer.
But then, exactly, how long does this event last? I mean, in real world we wouldn't worry about that, because our processors can process many operations per second, much faster than we can click, but what if I could click with a speed of light?
So, how exactly does KEY_DOWN is being timed?
Input events, like key-down, are buffered. When Windows receives the notification from the keyboard driver, it stores the event in the message queue. Which is owned by the thread that owns the foreground window. It will sit there however long it takes to be retrieved by your program. It needs to call GetMessage() or PeekMessage().
How long that takes is entirely unpredictable, it depends on how responsive the UI thread of the program is. Or in other words, how quickly it dispatches messages that it receives and calls GetMessage() again to retrieve the next messsage. A UI thread going comatose for a while is not unusual. Even on fast processors, a poorly written program that executes a dbase query that returns thousands of rows and waits for its result is common enough.
The buffering ensures that this doesn't cause a problem. No matter how fast the user types, his keystrokes cannot get lost.
Related
I'm implementing a Win32 Console text editor that has an internal message queue for passing around information about which areas to redraw, messages to/from plugins, etc. I prefer it to be single-threaded by default (if there's nothing happening that requires additional threads). I'm considering 2 message queue implementation strategies:
Use general-purpose queue and Win32 Event, so I can use WaitForMultipleObjectsEx to wait for internal messages and user input simultaneously, passing both console input handle and Event handle. In this case, text editor can live entirely within a single thread.
Use I/O Completion Port. In this case, text editor will need at least two threads, one of them calling GetQueuedCompletionStatus to get messages, and another reading user input and sending it to queue via PostQueuedCompletionStatus. The reason it that console input handle cannot be overlapped and WaitFor* functions don't accept Completion Port as a waitable handle, so it's not possible to wait on them simultaneously. Just like in the first setup, both threads don't waste CPU time when there's no input or events, but each keypress has to be passed from one thread to another via IOCP.
Which design is overall better?
Is performance and latency drawback from passing each keypress via IOCP significant for a text editor?
Performance and latency of IOCP is fine for your purpose. I wouldn’t however translate each key press into PostQueuedCompletionStatus. I’d rather PostQueuedCompletionStatus to enqueue multiple keypresses at once, whatever count you get from ReadConsoleInput.
I think performance difference is minimal, either one can be faster depending on the environment. And WaitForMultipleObjects is much easier to implement.
P.S. You sure you need message queue at all? Why don’t you process these redraw requests / plugin messages in whichever thread fired those, using e.g. critical section to guard (console output + your shared state)? If your congestion is low, like 100Hz messages + 15Hz keypresses, and you’ll process them fast, this will get you even lower latency than IOCP or any other queues: with low congestion or brief locking, critical sections don’t even switch to kernel for lock/unlock. This will also simplify design, your main thread will sleep on blocking ReadConsoleInput() call, no need to WaitFor.. before that.
Modbus is a a request and response type serial communication. Basically the master send out a request and one of the slave response.
I am modifing the code on a microcontroller which is a master unit on a modbus network. This unit also has a small dot-matrix LCD and some buttons for user interface. The microcontroller is running at 16MHz.
The problem is after the master unit send out a request, it does not know when the slave response, so it may need to wait for a relatively long time. However as this unit has buttons and LCD, it can not wait at a point for too long because the user will feel lag when he pressed a button. The original code is using a RTOS. It seperate the user interface task and the serial communication tasks so it has no problem. Now I need to change it to non-RTOS code. I have implemented a system tick timer which will interrupt at each 1ms. What is the proper (or common) way to do that?
It is possible to do quite a lot with just a single task, especially if you have interrupts. The intermediate position between a single very simple task and an RTOS is a cyclic executive. See http://www3.nd.edu/~cpoellab/teaching/cse40463/slides10.pdf for a brief overview of the spectrum of functionality from a cyclic executive up to a fully preemptive multitasking operating system. You will find much more if you search on this phrase and related phrases, including very sophisticated schemes for making sure that the system never misses its deadlines. If you are an aircraft flight control system, forgetting to check the aircraft pitch angle every X ms can cause problems elsewhere :-)
One way to rewrite code which is naturally multi-threaded is to maintain a model of the state of the system, such as a collection of objects each representing a modbus connection, indexed by a connection id. Then write a routine for every sort of event that can happen, including the arrival of a clock interrupt. When that event happens these routines typically work out which connection is involved, retrieve it from the main collection (or create it from scratch and enter it there if necessary) do the work associated with that particular sort of event, and then return.
It is often convenient to keep a queue of future events, indexed by time, and to have a routine that creates an object representing something to be done at some future time (such as calling a method to check for the expiration of a timeout) and puts this object on the queue.
You need to worry about interrupt processing getting called halfway through an event service routine. One way to deal with this is to lock out interrupts when that could cause a problem. Another way is to have the interrupt routine do nothing more than put an object on a queue that something else will check for later, or just set a flag. Then you need only lock out interrupts when you are checking for items on the queue and removing them.
A number of communications protocols are implemented in this way. Even in a true multitasking operating system you very often don't want to have to create a new thread every time you need to create a new connection. The two main problems with this is that the code is less clear than code which has a thread per object, because stuff that naturally goes together is chopped up into loads of event service events, and if any of the event service methods burn significant amounts of cpu, the system will stall because nothing else will happen when this is going on.
I am fixing a MFC applivation written in C++. It is a GUI and it communicates with an external module connected to the PC via USB.
I want to avoid using a separate thread. Is there a way I can add things to the mainloop so that it runs continously rather than being event based?
I want the mainloop to make a call to a function runCommStack() in every loop.
Some possible approaches:
You can use CWnd::SetTimer to set a timer.
You can override CWinApp::OnIdle (called by CWinApp::Run).
You can override CWinApp:Run, copying and modifying the original MFC's CWinApp:Run. This definitely is not the easiest solution.
You can create a background thread.
It depends on the requirements of runCommStack(). Is this function running long times? Then you probably won't want to run it in the GUI thread. Does runCommStack need to get called every n milliseconds? Then it might also be better to run it in it's own thread. In other cases you can just use the timer or OnIdle approach.
Regarding solution 1: as Tim pointed out WM_TIMER messages are low priority messages and will not be passed to the application while other higher-priority messages are in the message queue. See also Determine priority of a window message.
With solution 2 you should remind that OnIdle will only be called if no window message is available. So this is quite the same as solution 1 (in fact a little worse).
Also keep in mind that solutions 2 and 3 might result in your runCommStack not getting called if a dialog's DoModal() is called or if a message box is displayed. This is because during MessageBox() the control is not returned to CWinApp::Run().
I'ld implement solution 1 or 4.
When there are no messages (like keys, mouse, repaint) arriving the main loop suspends the program, waiting for the next message. So the idea of adding a call to the main loop will give you very erratic operation. For example, minimizing your window will stop all the USB communication.
Using SetTimer and calling your runCommStack function in a WM_TIMER handler will probably be much more satisfactory.
You can use idle processing with CWinApp::OnIdle; this will work if reading your USB device takes only a short amount of time, otherwise the user interface will be blocked during long reads.
But using a separate thread is definitely a better method.
I have a DirectShow filter written in Delphi 6 using the DSPACK component library. It is a push source video filter that receives its source frames from an external cooperating process that I also wrote.
When the worker thread that calls my Filters' FillBuffer() call is created and ran, when the graph starts up, the first thing I do from that worker thread is create a hidden window using AllocateHWND() to process WM_COPYDATA messages that contain the externally generated frames. Right before the thread is destroyed I destroy the hidden window. In other words the hidden window is created and destroyed in the execution context of the worker thread that calls FillBuffer(). My intention is to let FillBuffer() block as it waits for a WM_COPYDATA or a WM_QUIT message. The external cooperating process will submit frames to my filter using a WM_COPYDATA message and the handle to my hidden windows' WndProcc(). I will post a WM_QUIT message in my override of the pin's Inactive() method (thanks for that tip #RomanR), to unblock the FillBuffer() call before the filter is shut down.
My question is, is it safe to call PeekMessage() or GetMessage() from the FillBuffer() call given this scenario? Or are there potential pitfalls that may arise from this occurring in the context of a DirectShow graph executing? Also, do you see any flaws in my overall approach here that I need to consider?
Safe, but not so reasonable too. FillBuffer is being called on a background worker thread which is typically have no windows on it. It would be perhaps only your window which you are going to implement message loop for. And the window is here only for the purpose of receiving WM_COPYDATA messages. It sounds like it can work out, but you would perhaps could do much easier without helper windows by passing your data between applications via named file mappings and events. In case of video (you have audio, right?) you would also be able to appreciate a smaller performance overhead.
I wish to launch a separate thread for handling window messages (via a blocking GetMessage loop), but still create the windows in the initial thread, afterward.
Within the separate thread, as soon as it launches, I am calling PeekMessage with PM_NOREMOVE to ensure a message queue exists (is this necessary?), followed by..
AttachThreadInput(initial thread id,GetCurrentThreadId(),true)
..before finally entering the message loop
I am not yet using a mutex or cs to ensure this is happening in time, but am merely using a Sleep statement in my initial thread for the sake of simplicity.
Regardless, window messages do not appear to be intercepted by the separate thread.
I am a little unsure as to whether I am doing this correctly, and would appreciate any possible guidance. Both threads are in the same process
Thank you all
That's not what AttachThreadInput does. Even after you attach your input queue to another thread, Windows still have thread affinity. Messages in the queue for a given window can only be removed from the queue by that window's thread.
What AttachTheadInput does is to make two threads share an input queue. This allows them to query information about the input state and know that the other thread will get the same answer for the same query. For instance, one thread could call GetAsyncKeyState and know that the answer reflected the key state for the other thread.
It allows two or more threads to have the same relationship to the input queue and each other as processes had in Windows 3x. This is the reason that this API exists; so that complex multiprocess applications could be ported from Win 3x to Win95/WinNT.
It seems the best way to instigate window creation from the main thread, while having messages for them handled in a separate, looping thread is to use a custom message, that can be sent to the separate thread - Thus allowing it to create the window, but still allowing that action to be invoked from the initial thread:
1) Allocate a custom message, and create a structure to hold the window initialisation parameters:
message_create_window = WM_USER + 0;
class Message_create_window{
Message_create_window(...);
};
2) Instead of calling CreateWindow(Ex), use something similiar to the following, passing in the relavant window creation parameters:
PostThreadMessage(
thread.id,
message_create_window,
new Message_create_window(...),
0
);
3) Handle the custom message in the message pump of your ui handling thread, extract the creation parameters, & free afterwards:
MSG msg;
GetMessage(&msg,0,0,0);
...
switch(msg->message){
...
case message_create_window:{
Message_create_window *data=msg->wParam;
CreateWindowEx(data->...);
delete data;
}break;
...
This does, however, have the following side-effects:
The window will be created asynchronously. If it is required that the initial thread block until the window is created (or, indeed, that the window's existence can ever be asserted) then a thread synchronisation tool must be used (such as an event)
Care should be taken when interacting with the window (it is a multithreaded application, after all)
If there are any major holes in this answer, or this seems like a terrible approach, please correct me.
(This is still my question, & I am trying to find the best way to accomplish this)