>>>>> Todd Vierling writes:

Todd> Exactly how much overhead are we looking at between longcall and shortcall
Todd> here?  I just want to know how many CPU cycles we're anticipating for the
Todd> difference in calling convention, and whether it could be worth doing
Todd> -mlong-call universally, or just as a documented prerequisite for
Todd> LKM-capable kernels.  You could have something like:

	It depends on how the code is written and how efficient the
compiler is.  It means calling through a function pointer which means
finding the address of the symbol, moving that to the counter or link
register, and jumping through that register.  Loading the symbol takes at
least two cycles and moving the result from a GPR to a branchable register
takes another five cycles.

David