[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAM_iQpV6YQ6nh_Y0f3BDgMheOic3bPiqQ1oNHR+saxxE7nBuoA@mail.gmail.com>
Date: Mon, 9 Jan 2017 09:59:34 -0800
From: Cong Wang <xiyou.wangcong@...il.com>
To: Eric Dumazet <eric.dumazet@...il.com>
Cc: Jesper Dangaard Brouer <brouer@...hat.com>,
Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: Re: [net-next PATCH 1/3] Revert "icmp: avoid allocating large struct
on stack"
On Mon, Jan 9, 2017 at 9:50 AM, Eric Dumazet <eric.dumazet@...il.com> wrote:
> On Mon, 2017-01-09 at 09:42 -0800, Cong Wang wrote:
>> On Mon, Jan 9, 2017 at 7:04 AM, Jesper Dangaard Brouer
>> <brouer@...hat.com> wrote:
>> > This reverts commit 9a99d4a50cb8 ("icmp: avoid allocating large struct
>> > on stack"), because struct icmp_bxm no really a large struct, and
>> > allocating and free of this small 112 bytes hurts performance.
>>
>> The original commit fixes a warning for large stack usage, icmp_send()
>> is deep in the call stack.
>>
>> Your optimization for a slow path makes no sense to me.
>
> Do you have the stack trace of this event ?
>
> Even Linus allowed vmalloc() kernel stacks, while it certainly was an
> heresy 10 years ago.
>
> I doubt it makes a difference trying to save 104 bytes of kernel stack.
I think you should have known this, quote from Eric Dumazet
(hopefully the same one):
On Fri, 2013-05-31 at 22:22 -0700, Eric Dumazet wrote:
Strange, I posted a patch like that some days ago.
which is from: https://patchwork.ozlabs.org/patch/248051/
Facepalm...
Powered by blists - more mailing lists