Yes, there is additional overhead when IPC is a module. If the module
is loaded, the overhead is one test and one function call. If the
module isn't loaded, request_module is invoked and the overhead is
astronomical, but that can't happen in your inner loop.
The overhead when the module is loaded could be reduced by not
creating another stack frame. I don't know if GCC has extensions for
that; if it doesn't, the hooks would need to be in assembler. I wrote
some lmbench tests, which indicate that the overhead when the module
is loaded is lost in the noise, so I'm inclined not to bother.
zw
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/