Re: [PATCH v3 2/9] rbtree: add __rb_change_child() helper function

From: Daniel Santos
Date: Wed Sep 26 2012 - 18:34:05 EST


Sorry to resurrect the dead here, but I'm playing catch-up and this
looks important.

On 08/20/2012 05:17 PM, Andrew Morton wrote:
> I'm inclined to agree with Peter here - "inline" is now a vague,
> pathetic and useless thing. The problem is that the reader just
> doesn't *know* whether or not the writer really wanted it to be
> inlined.
>
> If we have carefully made a decision to inline a function, we should
> (now) use __always_inline.
Are we all aware here that __always_inline (a.k.a.
"__attribute__((always_inline))") just means "inline even when not
optimizing"? This appears to be a very common misunderstanding (unless
the gcc docs are wrong, see
http://gcc.gnu.org/onlinedocs/gcc/Function-Attributes.html#index-g_t_0040code_007bflatten_007d-function-attribute-2512).

If you want to *force* gcc to inline a function (when inlining is
enabled), you can currently only do it from the calling function by
adding the |flatten attribute to it, which I have proposed adding here:
https://lkml.org/lkml/2012/9/25/643.

Thus, all of the __always_inline markings we have in the kernel only
affect unoptimized builds (and maybe -O1?). If we need this feature
(and I think it would be darned handy!) we'll have to work on gcc to get it.

Daniel
|
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/