I won't apply this version, please send v2 with the changes below.
On Fri, 06 Nov 2026 05:02:08 +0200, Eva Kurchatova
<[email protected]> wrote:
> Paolo Abeni (8):
> ms/mptcp: fix retransmission loop when csum is enabled
> ms/mptcp: move the retrans loop to a separate helper
> ms/mptcp: move the stale logic out of retrans scheduler
> ms/mptcp: let the retrans scheduler do its job
> ms/mptcp: explicitly drop over memory limits
> ms/mptcp: implemented OoO queue pruning
> ms/mptcp: being below memory limit is a likely() condition
> ms/mptcp: avoid pruning for OoW data
The backports themselves match mainline. What to change for v2:
1. Add mainline 8ab24fdebc369 ("mptcp: close TOCTOU race while
computing rcv_wnd") and da23be77e1292 ("mptcp: allow subflow rcv
wnd to shrink") before patch 5. Both are Cc stable and come from
the same 7.1-rc7 batch as patch 1.
2. Cover letter: say what the series fixes (endless retransmission
with csum, receive stalls under memory pressure), which patches are
only prerequisites (2-4), what was left out of the mainline series
and why, and how it was tested.
3. Patch 3: fix the rebase note, mainline also moves only "stale" out
of the bitfield.
4. Patch 6: name the two skipped mainline commits in the rebase note.
Separate patches, not needed for v2; I created a task for each:
1. Mainline 7538fa33bb2f6 ("mptcp: fix subflow bitfield misuse").
Our tree still flips close_event_done under the msk lock only,
in the same word as bits written under the subflow lock.
https://virtuozzo.atlassian.net/browse/VSTOR-154927
2. Mainline e2ab913f68c7d ("mptcp: do not reschedule the RTX timer
for fallback sockets"). After a fallback to TCP the MPTCP
retransmit timer keeps being rearmed and fires as a no-op.
https://virtuozzo.atlassian.net/browse/VSTOR-154928
--
Konstantin Khorenko,
Virtuozzo Linux Kernel Team
_______________________________________________
Devel mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/devel