Hi all,

Currently detach support is skipped on MinGW (and mingw32ce),
because there is no waitpid on Windows.  This patch enables it
by implementing the waiting for the inferior to exit in a new
target_op::join.

One thing that annoys me, is that requesting for detach kills the
inferior if detaching isn't supported.  The patch makes gdbserver
return an error (E packet) to gdb, so gdb knows it shouldn't
stop debugging.  The user can then kill the inferior is he wants.

What do you think of this behavior?

Is the patch OK?

Cheers,
Pedro Alves