Re: [Valgrind-users] cachegrind annotation of inline functions

[Nicholas N. <n.n...@gm...>]

On Fri, May 29, 2009 at 7:16 AM, Brian J. Miller <mil...@ll...> wrote:
>
> I'm using 3.4.1 and am wondering if the location of the
> annotation indicates whether the compiler actually inlined
> a function declared inline.  I have some inline functions
> whose annotated cache data appears in the .hh file while
> other inline functions of the same class have their annotated
> data appear in the source file where they are called.
>
> Does this mean that the latter function was inlined while
> the former was not?

It's possible.  Cachegrind relies entirely on the debug info to
annotate the code, but it's possible that the debug info is flawed.

In order to truly tell if inlining has occurred I'd recommend looking
at the assembly code.

Nick