shevek_closure(3) [debian man page]
shevek::closure(3) Library Functions Manual shevek::closure(3) NAME
shevek::closure - Block and resume without blocking the main loop. SYNOPSIS
#include <closure.hh> Inherits shevek::refbase. Public Member Functions bool empty () const Check if the closure is empty. void set_function (sigc::slot0< void > func, bool run=true, sigc::slot0< void > cb=sigc::slot0< void >()) Set running function on an empty closure. ~closure () Destructor. void wake () Continue running the closure. Static Public Member Functions static Glib::RefPtr< closure > create () Create a new closure. static void block () Sleep, returning control to the caller until awoken. Detailed Description Block and resume without blocking the main loop. Closures allow blocking and resuming the main loop. They are implemented with threads, which means that they don't work well with multi- threaded programs. If a function wants to be able to block using a closure, it must be called using closure(). It (or any function it calls) can then suspend by calling closure::block(). It can be awoken again by calling closure::wake() on the closure object. Member Function Documentation static void shevek::closure::block () [static] Sleep, returning control to the caller until awoken. This function puts the current closure to sleep. It will continue to run when awoken with wake(). It can also be destroyed. This function uses a global variable to know which is the current closure, so it can be called without an object, as closure::block (); . static Glib::RefPtr<closure> shevek::closure::create () [inline, static] Create a new closure. Create a new closure. It will be empty initially. bool shevek::closure::empty () const [inline] Check if the closure is empty. Check if the closure is empty. If it is, set_function() can be called. void shevek::closure::set_function (sigc::slot0< void >func, boolrun = true, sigc::slot0< void >cb = sigc::slot0< void >()) Set running function on an empty closure. Set running function. The closure must be empty when this is called. When the function exits, the closure returns to the empty state, and the callback is called, if given. void shevek::closure::wake () Continue running the closure. Wake a closure. It is an error to wake a closure which isn't blocking (in particular also the currently running closure). Author Generated automatically by Doxygen for libshevek from the source code. libshevek Fri May 11 2012 shevek::closure(3)
Check Out this Related Man Page
ffi_prep_closure(3) BSD Library Functions Manual ffi_prep_closure(3) NAME
ffi_prep_closure -- Prepare a ffi_closure for execution. SYNOPSIS
#include <ffi/ffi.h> ffi_status ffi_prep_closure(ffi_closure *closure, ffi_cif *cif, void (*fun)(ffi_cif*,void*,void**,void*), void *user_data); DESCRIPTION
closure is prepared to execute fun. cif contains information describing the data types, sizes and alignments of the arguments to and return value from the function that will be called from fun, and must be initialized with ffi_prep_cif before it is used with ffi_prep_closure. user_data may point to additional data to be used in fun. If no additional data is needed, user_data may be NULL. When closure is invoked, fun is called with cif, an array of pointers to arguments, a pointer to a return value, and user_data. Some architectures do not allow the execution of data by default. In such cases, it is necessary to manually alter the permissions of the page that contains closure prior to its execution. RETURN VALUES
Upon successful completion, ffi_prep_closure returns FFI_OK. If the ABI specified in cif does not refer to a valid ABI, FFI_BAD_ABI will be returned. Available ABIs are defined in <ffi/ppc-ffitarget.h> and <ffi/x86-ffitarget.h>. EXAMPLES
#define MACOSX // for fficonfig.h on Darwin #include <ffi/ffi.h> #include <sys/mman.h> // for mmap() unsigned char foo(unsigned int, float); static void foo_closure(ffi_cif*, void*, void**, void*); int main(int argc, const char **argv) { ffi_cif cif; ffi_closure *closure; ffi_type *arg_types[2]; ffi_arg result; ffi_status status; // Specify the data type of each argument. Available types are defined // in <ffi/ffi.h>. arg_types[0] = &ffi_type_uint; arg_types[1] = &ffi_type_float; // Allocate a page to hold the closure with read and write permissions. if ((closure = mmap(NULL, sizeof(ffi_closure), PROT_READ | PROT_WRITE, MAP_ANON | MAP_PRIVATE, -1, 0)) == (void*)-1) { // Check errno and handle the error. } // Prepare the ffi_cif structure. if ((status = ffi_prep_cif(&cif, FFI_DEFAULT_ABI, 2, &ffi_type_uint8, arg_types)) != FFI_OK) { // Handle the ffi_status error. } // Prepare the ffi_closure structure. if ((status = ffi_prep_closure(closure, &cif, foo_closure, NULL)) != FFI_OK) { // Handle the ffi_status error. } // Ensure that the closure will execute on all architectures. if (mprotect(closure, sizeof(closure), PROT_READ | PROT_EXEC) == -1) { // Check errno and handle the error. } // The closure is now ready to be executed, and can be saved for later // execution if desired. // Invoke the closure. result = ((unsigned char(*)(float, unsigned int))closure)(42, 5.1); // Free the memory associated with the closure. if (munmap(closure, sizeof(closure)) == -1) { // Check errno and handle the error. } return 0; } // Invoking the closure transfers control to this function. static void foo_closure(ffi_cif* cif, void* result, void** args, void* userdata) { // Access the arguments to be sent to foo(). float arg1 = *(float*)args[0]; unsigned int arg2 = *(unsigned int*)args[1]; // Call foo() and save its return value. unsigned char ret_val = foo(arg1, arg2); // Copy the returned value into result. Because the return value of foo() // is smaller than sizeof(long), typecast it to ffi_arg. Use ffi_sarg // instead for signed types. *(ffi_arg*)result = (ffi_arg)ret_val; } // The closed-over function. unsigned char foo(unsigned int x, float y) { unsigned char result = x - y; return result; } SEE ALSO
ffi(3), ffi_prep_cif(3), mmap(2), munmap(2), mprotect(2) Darwin July 20, 2007 Darwin