On 29/11/19 - 21:13:38, Alan Ford wrote:
> On 28 Nov 2019, at 19:49, Christoph Paasch
>> On Nov 28, 2019, at 8:16 AM, Alan Ford <alan.ford(a)gmail.com
>>> On 27 Nov 2019, at 19:29, Christoph Paasch <cpaasch(a)apple.com
>>> Section 3.3.1, page 32 & 33, "A data sequence mapping does not
>>> This paragraph states that it is permissive to send a mapping in advance.
Late-mapping is specified a bit higher around the sentence
>>> "Implementations MAY hold onto such unmapped data for a short while in
the expectation that a mapping will arrive shortly"
>>> This kind of early/late mapping announcements are difficult to handle in an
implementation. The Linux Kernel implementation of multipath-tcp.org
has always disallowed such kind of mappings. Meaning,
whenever a DSS-option is received such that the range specified by the relative
subflow-sequence number in the DSS-option and the data-length does not (partially) cover
the TCP sequence number of the packet itself, the subflow will be killed with a TCP-RST.
The problem around handling such early/late mappings is that it is unclear for how long
the stack needs to remember these mappings (in the early-mapping case), or for how long he
needs to hold on to the data (in the late-mapping case).
>>> We thus suggest to change this to the following:
>>> Whenever a DSS-option is received on a packet such that the mapping of the
subflow-sequence space does not partially cover the TCP-sequence number of the packet
itself, the host MUST discard this mapping and MAY destroy the subflow with a TCP-RST. It
should be noted that a DATA_FIN that does not come with data has a relative
subflow-sequence number of 0 and thus should be handled separately.
>> This one I do have an issue with:
>> - It is a technical change
>> - Wording to this effect has been in the document since pretty much the
>> - It is a MAY which might as well say “there is no guarantee this would work”
> The problem with the MAY is that the sender can't really know if the receiver
accepts it (more regarding this below)
>> Most importantly, the replacement text seems not to address this issue at all.
If I read it correctly, it says that the data sequence mapping option MUST partially cover
the subflow sequence space of the packet itself. But that has nothing to do with late or
early mapping, both could partially cover the subflow sequence space and preceding data.
>> Can you clarify exactly what you want to permit and prevent, here?
> Let me try to clarify what exactly we mean with early/late mapping so that we are
all on the same terms here:
> Early mapping:
> A TCP-segment with sequence-number 1 holds a DSS-option with subflow-sequence number
1001 and data-length 100. This means we need to allocate space to store this DSS-option so
that when the TCP-segment with seqno 1001 arrives we can know the mapping. There may be
coming more of these DSS-options which all need to be stored in allocated memory. It is
unclear what the limit to this is and there is no way to communicate this limit to the
I don’t think we have ever intended to support a mapping like this. If the text is not
clear here then yes, we might have an issue.
Yes, I do think that the text is not very clear on that - we should clarify
We intend only to support: A TCP segment with sequence-number 1 holds
a DSS option for SSN 1 and length 10000 (so multiple segments in the future).
Or, slightly more convoluted, a TCP segment with SN 100, length 100,
which holds a DSS option for 151-250, and bytes 50-150 were already covered by a previous
I do not believe we have ever intended mappings on data segments
where the segments do not include any of the mapped data. We did, however, intend to
support mappings on pure ACKs in order to avoid any option space limits.
Mappings on a pure ACK are an unlikely use-case. The problem is that in the
end the mapping needs to reliably make it to the receiver as otherwise he
needs to throw the data away. Combining it with data implicitly makes it
In case of option-space limits, it is better to send the options that do
consume a lot of space (ADD_ADDR,...) on the pure ACK.
> Late mapping:
> The receiver receives data without DSS-options with TCP-sequence 1 to 1001. The
corresponding DSS-option however arrives with the TCP-segment with seqno 2001. Here, the
receiver needs to hold on to this data, waiting for the TCP-segment with the DSS-option.
At one point the receiver needs to drop the data due to memory limits. Again, the sender
has no way for knowing what this limit is.
So this is slightly more problematic, and this is what the MAY in the text is designed to
I believe we had intended to support a situation where you could have a segment 1-1000
without a DSS option, and 1001-2000 also without, and then 2001-3000 with the option then
providing a mapping for 1 to 2001 or higher.
I think this scenario is fine to some extend with a slight change that the
provided mapping should be for 1 to 2002 or higher (thus, including the byte
of the 2001-3000 segment).
This would allow a sender to start pushing out data before it knew
what a mapping might look like.
It does, however, seem an unlikely situation but you as a receiver could of course reduce
the subflow window size in order to limit buffering. The text does recognise the memory
issue and point out that this won’t be DATA ACKed and as such a sender should soon realise
and retransmit on a separate subflow and then this subflow may eventually be closed.
Yes, it seems like an unlikerly situation and I'm not sure about the
use-case for doing this.
Another situation would involve two mappings on the same TCP segment.
If the first 50 bytes of a segment are covered by the mapping provided in that segment,
but there are then also another 50 bytes, then the mapping can’t be provided until the
Yes, two mappings on a single packet are problematic simply because of the
lack of option-space.
I guess my initial thought here is that this was intended to cover a
number of corner cases but if it does not work for you then there are several compliant
ways of dealing with it.
> When the DSS-option comes together with the corresponding TCP-sequence it is
straight-forward to store it together with the data. There are no issues with
memory-allocation,... as all of this is accounted together with the announced window (yes,
the memory is not counted against the window, but the receiver can foresee the DSS-option
overhead when computing what window he should announce).
> When a receiver gets data without a DSS-option, he can store it for up to 64KB of
data as that is the maximum data-level length and the last segment of the 64KB-train could
be holding the DSS-option. After that he has to drop the data.
> When the mapping partially covers the segment it also isn't a problem as the
unmapped part can safely be dropped and the mapped part can be passed on to the
> All of this does not imply that every segment of a mapping needs to hold the
DSS-option. Just one of them needs to have it.
That last statement aligns with what was intended in the text. But are you really saying
that, or are you saying that the first one needs to have it? Because that would be a
No, the first one does not need to have it. Just one of them.