I won't apply this version, please send v3 with the changes below.
On Mon, 16 Nov 2026 05:49:07 +0200, Eva Kurchatova
<[email protected]> wrote:
> Changes in v2:
> - Add "ms/mptcp: close TOCTOU race while computing rcv_wnd" and
> "ms/mptcp: allow subflow rcv wnd to shrink before the drop over memory
> limits".
The backports match mainline and the rebase notes are accurate. What
to change for v3:
1. Add mainline 2834f8edd74d5 ("mptcp: cleanup fallback dummy mapping
generation") before patch 7. Without it, a drop by patch 7 followed
by a fallback to TCP silently cuts bytes out of the stream.
2. Patch 6: say in a rebase note that mainline 0e24d17bd9668 ("tcp:
implement RFC 7323 window retraction receiver requirements"), which
it sits on in mainline, is not taken, as in 6.12.y.
3. Cover letter: item 2 of my v1 reply is still open. Say what the
series fixes (endless retransmission with csum, receive stalls
under memory pressure), which patches are only prerequisites, what
was left out of the mainline series and why, and how it was tested.
Also bring back the Jira link, name the base of the rebase, and
close the quote after "allow subflow rcv wnd to shrink".
Separate patches, not needed for v3; I created a task for each:
1. Already in the base: when sk_rmem_schedule() fails in
__mptcp_move_skb() (tcp_mem limit, memcg charge in a container),
TCP-acked data is dropped even on a fallback socket, and the next
bytes take its place in the stream. Mainline removed that drop with
9db5b3cec4ec ("mptcp: borrow forward memory from subflow").
https://virtuozzo.atlassian.net/browse/VSTOR-155321
2. Four Cc stable mptcp fixes this tree lacks: 42064de57fb83,
f3ef03357396d, 85c580b0d8590, 0981f90e1a05.
https://virtuozzo.atlassian.net/browse/VSTOR-155322
--
Konstantin Khorenko,
Virtuozzo Linux Kernel Team
_______________________________________________
Devel mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/devel