From: Steffen Klassert <steffen.klassert@secunet.com>
To: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Cc: David Miller <davem@davemloft.net>,
Paolo Abeni <pabeni@redhat.com>,
Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>,
Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>,
Network Development <netdev@vger.kernel.org>
Subject: Re: [PATCH net-next 3/4] net: Support GRO/GSO fraglist chaining.
Date: Mon, 20 Jan 2020 09:35:18 +0100 [thread overview]
Message-ID: <20200120083518.GL23018@gauss3.secunet.de> (raw)
In-Reply-To: <CA+FuTSeF06hJstQBH4eL4L3=yGdiizw_38BUheYyircW8E3cXg@mail.gmail.com>
On Wed, Jan 15, 2020 at 10:43:08AM -0500, Willem de Bruijn wrote:
> > > > Maybe we can be conservative here and do a full
> > > > __copy_skb_header for now. The initial version
> > > > does not necessarily need to be the most performant
> > > > version. We could try to identify the correct subset
> > > > of header fields later then.
> > >
> > > We should probably aim for the right set from the start. If you think
> > > this set is it, let's keep it.
> >
> > I'd prefer to do a full __copy_skb_header for now and think a bit
> > longer if that what I chose is really the correct subset.
>
> Ok
>
> > > > > > I had to set ip_summed to CHECKSUM_UNNECESSARY on GRO to
> > > > > > make sure the noone touches the checksum of the head
> > > > > > skb. Otherise netfilter etc. tries to touch the csum.
> > > > > >
> > > > > > Before chaining I make sure that ip_summed and csum_level is
> > > > > > the same for all chained skbs and here I restore the original
> > > > > > value from nskb.
> > > > >
> > > > > This is safe because the skb_gro_checksum_validate will have validated
> > > > > already on CHECKSUM_PARTIAL? What happens if there is decap or encap
> > > > > in the path? We cannot revert to CHECKSUM_PARTIAL after that, I
> > > > > imagine.
> > > >
> > > > Yes, the checksum is validated with skb_gro_checksum_validate. If the
> > > > packets are UDP encapsulated, they are segmented before decapsulation.
> > > > Original values are already restored. If an additional encapsulation
> > > > happens, the encap checksum will be calculated after segmentation.
> > > > Original values are restored before that.
> > >
> > > I was wondering more about additional other encapsulation protocols.
> > >
> > > >From a quick read, it seems like csum_level is associated only with
> > > CHECKSUM_UNNECESSARY.
> > >
> > > What if a device returns CHECKSUM_COMPLETE for packets with a tunnel
> > > that is decapsulated before forwarding. Say, just VLAN. That gets
> > > untagged in __netif_receive_skb_core with skb_vlan_untag calling
> > > skb_pull_rcsum. After segmentation the ip_summed is restored, with
> > > skb->csum still containing the unmodified csum that includes the VLAN
> > > tag?
> >
> > Hm, that could be really a problem. So setting CHECKSUM_UNNECESSARY
> > should be ok, but restoring the old values are not. Our checksum
> > magic is rather complex, it's hard to get it right for all possible
> > cases. Maybe we can just set CHECKSUM_UNNECESSARY for all packets
> > and keep this value after segmentation.
>
> Note that I'm not 100% sure that the issue can occur. But it seems likely.
>
> Yes, inverse CHECKSUM_UNNECESSARY conversion after verifying the checksum is
> probably the way to go. Inverse, because it is the opposite of
> __skb_gro_checksum_convert.
I'm not sure if I understand what you mean here. I'd do the following
for fraglist GRO in udp4_gro_complete:
if (skb->ip_summed == CHECKSUM_UNNECESSARY) {
if (skb->csum_level < SKB_MAX_CSUM_LEVEL)
skb->csum_level++;
} else {
skb->ip_summed = CHECKSUM_UNNECESSARY;
skb->csum_level = 0;
}
and then copy these values to the segments after segmentation.
next prev parent reply other threads:[~2020-01-20 8:35 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-18 13:34 [PATCH net-next 0/4] Support fraglist GRO/GSO Steffen Klassert
2019-12-18 13:34 ` [PATCH net-next 1/4] net: Add fraglist GRO/GSO feature flags Steffen Klassert
2019-12-18 16:02 ` Willem de Bruijn
2019-12-18 13:34 ` [PATCH net-next 2/4] net: Add a netdev software feature set that defaults to off Steffen Klassert
2019-12-18 16:02 ` Willem de Bruijn
2019-12-18 13:34 ` [PATCH net-next 3/4] net: Support GRO/GSO fraglist chaining Steffen Klassert
2019-12-18 16:02 ` Willem de Bruijn
2019-12-19 8:22 ` Steffen Klassert
2019-12-19 16:28 ` Willem de Bruijn
2020-01-13 8:51 ` Steffen Klassert
2020-01-13 16:21 ` Willem de Bruijn
2020-01-15 9:47 ` Steffen Klassert
2020-01-15 15:43 ` Willem de Bruijn
2020-01-20 8:35 ` Steffen Klassert [this message]
2020-01-20 16:35 ` Willem de Bruijn
2019-12-18 13:34 ` [PATCH net-next 4/4] udp: Support UDP fraglist GRO/GSO Steffen Klassert
2019-12-18 16:03 ` Willem de Bruijn
2019-12-19 8:26 ` Steffen Klassert
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=20200120083518.GL23018@gauss3.secunet.de \
--to=steffen.klassert@secunet.com \
--cc=davem@davemloft.net \
--cc=marcelo.leitner@gmail.com \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=subashab@codeaurora.org \
--cc=willemdebruijn.kernel@gmail.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;
as well as URLs for NNTP newsgroup(s).