I see we have a new "call_usermodehelper" routine now. It looks like the
end result is similer to exec_usermodehelper, except that you no longer
need to mess with creating kernel threads yourself.
Is call_usermodehelper now the officially blessed way for kernel code to
run something in user space? Perhaps exec_usermodehelper should become
private to kmod.c?
I also see that call_usermodehelper will call do_exit() if the exec fails,
while the path taken in request_module does not do that. Can both be
right?
Thanks,
jon
Jonathan Corbet
Executive editor, LWN.net
[email protected]