After setting the ioctl() command you're suggesting on my socket, I'm reading from it using sysread(), which behaves in a non-blocking way, returning the amount of bytes it has actually read. This overall works much faster than my looping approach, especially for longer messages.
Although this is probably off-topic, it's curious how the "magic value" of 0x8004667E is created. The "macro" for non-blocking IO in ioctl() is FIONBIO, and its calculation is done thus:
I wonder what is the rationale behind this ? Can someone explain how these things work ? Is it portable to just plunge the magic value in my code ?IOC_IN = 0x80000000h; FIONBIO = IOC_IN + 4 * 65536 + ORD ('f') * 256 + 126;