lkml.org 
[lkml]   [2005]   [Jan]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] micro optimization in kernel/params.c; don't call to_module_kobject before we really have to.
Jesper Juhl wrote:
> In kernel/params.c::module_attr_show and
> kernel/params.c::module_attr_store we call to_module_kobject() and save
> the result in a local variable right before a conditional statement that
> does not depend on the call to to_module_kobject() and may cause the
> function to return. If the function returns before we use the result of
> to_module_kobject() then the call is just a waste of time.
> The patch moves the call to to_module_kobject() down just before it is
> actually used, thus we save a call to the function in a few cases. I doubt
> this is in any way measurable, but I see no reason to not move the call -
> it should be an infinitesimal improvement with no ill sideeffects.
> Please consider applying.
>
>
> Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
>
> diff -up linux-2.6.11-rc2-bk7-orig/kernel/params.c linux-2.6.11-rc2-bk7/kernel/params.c
> --- linux-2.6.11-rc2-bk7-orig/kernel/params.c 2005-01-29 23:54:53.000000000 +0100
> +++ linux-2.6.11-rc2-bk7/kernel/params.c 2005-01-30 00:27:08.000000000 +0100
> @@ -625,11 +625,12 @@ static ssize_t module_attr_show(struct k
> int ret;
>
> attribute = to_module_attr(attr);
> - mk = to_module_kobject(kobj);
>
> if (!attribute->show)
> return -EPERM;
>
> + mk = to_module_kobject(kobj);
> +
> if (!try_module_get(mk->mod))
> return -ENODEV;
>
> @@ -649,11 +650,12 @@ static ssize_t module_attr_store(struct
> int ret;
>
> attribute = to_module_attr(attr);
> - mk = to_module_kobject(kobj);
>
> if (!attribute->store)
> return -EPERM;
>
> + mk = to_module_kobject(kobj);
> +
> if (!try_module_get(mk->mod))
> return -ENODEV;
>
>
>
>

I'd bet that the compiler already reorders the assignment since there is
no side effect to to_module_kobject(). Even with a patch like this you
can't control exactly when the assignment is done. There may not even
be an assignment, since mk is a constant offset of kobj.

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

\
 
 \ /
  Last update: 2005-03-22 14:10    [W:0.048 / U:0.432 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site