Re: [PATCH RFC V2 2/2] net: Optimize snmp stat aggregation by walking all the percpu data at once

From: Joe Perches
Date: Fri Aug 28 2015 - 20:35:35 EST


On Fri, 2015-08-28 at 17:06 -0700, Eric Dumazet wrote:
> On Fri, 2015-08-28 at 16:12 -0700, Joe Perches wrote:
> > Generally true. It's always difficult to know how much
> > stack has been consumed though and smaller stack frames
> > are generally better.
[]
> So for a _leaf_ function, it is better to declare an automatic variable,
> as you in fact reduce max stack depth.

That of course depends on what a "leaf" is and
whether or not any other function call in the
"leaf" consumes stack.

inet6_fill_ifla6_attrs does call other functions
(none of which has the stack frame size of k.alloc)

> Not only it uses less kernel stack, it is also way faster, as you avoid
> kmalloc()/kfree() overhead and reuse probably already hot cache lines in
> kernel stack.

yup.

You'll also never neglect to free stack like the
original RFC patch neglected to free the alloc.

cheers, Joe

--
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/