I won't apply this version, please send v3 with the changes below.
On Mon, 16 Nov 2026 05:49:14 +0200, Eva Kurchatova
<[email protected]> wrote:
> Rebase note: this tree has no MPTCP-level backlog, its receive path
> moves the skbs from the subflow queue straight to the msk. The drop goes
> into __mptcp_move_skb(), and the loops that move the skbs stop over the
Item 1 of the cover letter reply: put mainline 2834f8edd74d5 ("mptcp:
cleanup fallback dummy mapping generation") before this patch.
An in-sequence skb dropped here has already been acked by TCP; only
the MPTCP retransmission brings it back. If the connection then falls
back to TCP (one subflow, empty OoO queue, a segment without a DSS
mapping), subflow_check_data_avail() builds the dummy mapping from
msk->ack_seq. The next bytes get the DSN of the dropped skb, and the
reader sees a stream with bytes missing and no error.
2834f8edd74d5 builds that mapping from the TCP sequence number, so the
bytes after the hole keep their own DSN and never reach the reader in
the wrong place. Mainline had it before it started to drop here, and
everything it needs (iasn, __mptcp_expand_seq()) is in this tree.
--
Konstantin Khorenko,
Virtuozzo Linux Kernel Team
_______________________________________________
Devel mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/devel