Netdev List
 help / color / mirror / Atom feed
From: "Eilon Greenstein" <eilong@broadcom.com>
To: "Michal Schmidt" <mschmidt@redhat.com>,
	"Eric Dumazet" <eric.dumazet@gmail.com>
Cc: netdev@vger.kernel.org, "Eric Dumazet" <edumazet@google.com>,
	"Yaniv Rosner" <yanivr@broadcom.com>,
	"Yuval Mintz" <yuvalmin@broadcom.com>,
	"Merav Sicron" <meravs@broadcom.com>,
	"Robert Evans" <evansr@google.com>,
	"Tom Herbert" <therbert@google.com>,
	"Willem de Bruijn" <willemb@google.com>,
	"David Miller" <davem@davemloft.net>,
	"Havard Skinnemoen" <hskinnemoen@google.com>
Subject: Re: [PATCH] bnx2x: fix rx checksum validation for IPv6
Date: Fri, 14 Sep 2012 20:39:06 +0300	[thread overview]
Message-ID: <1347644346.4491.1.camel@lb-tlvb-eilong.il.broadcom.com> (raw)
In-Reply-To: <1347600028.23482.1.camel@lb-tlvb-eilong.il.broadcom.com>

On Fri, 2012-09-14 at 08:20 +0300, Eilon Greenstein wrote:
> On Fri, 2012-09-14 at 01:14 +0200, Eric Dumazet wrote:
> > On Fri, 2012-09-14 at 00:59 +0200, Michal Schmidt wrote:
> > > Commit d6cb3e41 "bnx2x: fix checksum validation" caused a performance
> > > regression for IPv6. Rx checksum offload does not work. IPv6 packets
> > > are passed to the stack with CHECKSUM_NONE.
> > > 
> > > The hardware obviously cannot perform IP checksum validation for IPv6,
> > > because there is no checksum in the IPv6 header. This should not prevent
> > > us from setting CHECKSUM_UNNECESSARY.
> > > 
> > > Tested on BCM57711.
> > > 
> > > Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
> > > ---
> > >  drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 12 +++++++-----
> > >  1 file changed, 7 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
> > > index af20c6e..e8e97a7 100644
> > > --- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
> > > +++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
> > > @@ -662,14 +662,16 @@ void bnx2x_csum_validate(struct sk_buff *skb, union eth_rx_cqe *cqe,
> > >  				 struct bnx2x_fastpath *fp,
> > >  				 struct bnx2x_eth_q_stats *qstats)
> > >  {
> > > -	/* Do nothing if no IP/L4 csum validation was done */
> > > -
> > > +	/* Do nothing if no L4 csum validation was done.
> > > +	 * We do not check whether IP csum was validated. For IPv4 we assume
> > > +	 * that if the card got as far as validating the L4 csum, it also
> > > +	 * validated the IP csum. IPv6 has no IP csum.
> > > +	 */
> > >  	if (cqe->fast_path_cqe.status_flags &
> > > -	    (ETH_FAST_PATH_RX_CQE_IP_XSUM_NO_VALIDATION_FLG |
> > > -	     ETH_FAST_PATH_RX_CQE_L4_XSUM_NO_VALIDATION_FLG))
> > > +	    ETH_FAST_PATH_RX_CQE_L4_XSUM_NO_VALIDATION_FLG)
> > >  		return;
> > >  
> > > -	/* If both IP/L4 validation were done, check if an error was found. */
> > > +	/* If L4 validation was done, check if an error was found. */
> > >  
> > >  	if (cqe->fast_path_cqe.type_error_flags &
> > >  	    (ETH_FAST_PATH_RX_CQE_IP_BAD_XSUM_FLG |
> > 
> > Thanks for fixing this bug !
> > 
> > Acked-by: Eric Dumazet <edumazet@google.com>
> 
> Indeed - thanks Michal!
> 
> Acked-by: Eilon Greenstein <eilong@broadcom.com>
> 

Just in case someone will look up this thread in the future, I’m adding
some details: the comment is almost right, the HW does not verify the
IPv6 header in case it contains extension headers  this is why this
patch is required (though some users with IPv6 that does not use any
extension headers will not see the issue).

In other words: Eric  this is how we both missed it…

  reply	other threads:[~2012-09-14 17:40 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-09-13 22:59 [PATCH] bnx2x: fix rx checksum validation for IPv6 Michal Schmidt
2012-09-13 23:14 ` Eric Dumazet
2012-09-14  5:20   ` Eilon Greenstein
2012-09-14 17:39     ` Eilon Greenstein [this message]
2012-09-18 20:17   ` David Miller

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1347644346.4491.1.camel@lb-tlvb-eilong.il.broadcom.com \
    --to=eilong@broadcom.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=eric.dumazet@gmail.com \
    --cc=evansr@google.com \
    --cc=hskinnemoen@google.com \
    --cc=meravs@broadcom.com \
    --cc=mschmidt@redhat.com \
    --cc=netdev@vger.kernel.org \
    --cc=therbert@google.com \
    --cc=willemb@google.com \
    --cc=yanivr@broadcom.com \
    --cc=yuvalmin@broadcom.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox