Transfer of synchronized data from 16-bit code to a 32-bit process

Abstract

The 16-bit process receives synchronized data from a hooked data function. As memory becomes available in a shared memory buffer, the 16-bit process writes the synchronized data to the buffer. The 16-bit process then signals the 32-bit process that synchronized data is ready to be transmitted over the network. The 32-bit process reads the synchronized data, stores it in a send buffer to free the shared memory buffer, then signals the 16-bit process that the shared memory buffer has been read. The 32-bit process then sends the synchronized data out over the network. In this way, a 'thunk' that improperly releases the mutual exclusion semaphore is avoided. There can be more than one shared memory buffer, to allow the 16-bit process and the 32-bit process to read and write synchronized data to the shared memory buffers at the same time. The 16-bit process can use a callback function to ensure that data does not wait in the shared memory buffers for too long before it is ready by the 32-bit process and sent out over the network.