netdev - Re: GRO support in the bridge

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <20110810080831.05de9a5f@nehalam.ftrdhcpuser.net>
Date:	Wed, 10 Aug 2011 08:08:31 -0700
From:	Stephen Hemminger <shemminger@...ux-foundation.org>
To:	Sedat Cakir <cakir_sedat@...oo.com>
Cc:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: GRO support in the bridge

On Wed, 10 Aug 2011 05:29:15 -0700 (PDT)
Sedat Cakir <cakir_sedat@...oo.com> wrote:

> Hi, 
> 
> 
> 
> I am trying to understand why GRO is NOT enabled on the bridge. Is there a special reason for it? Reading the code in net/core/dev.c:2817 (__netif_receieve_skb) which is what I think the bridge is calling because that is what is passed in at br_input.c:br_pass_frame_up():24. I am wondering if this can be modified to use one of the _gro_ receive functions and if so if that will positively impact performance? Do you have any idea? 
> 
> Appreciate your help.Thanks,
> Sedat

The reason is that GRO is related to NAPI.
GRO works by coalescing frames received in one NAPI interval
of work.  The bridge itself has no receive irq or NAPI and is upcalled
from the device receive handler. There is no expectation or
requirement that devices in a bridge use NAPI. Therefore it
is not possible or safe for bridge pseudo-device to use GRO.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html