From: Mina Almasry <almasrymina@google.com>
To: netdev@vger.kernel.org
Cc: Mina Almasry <almasrymina@google.com>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@kernel.org>,
Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
Simon Horman <horms@kernel.org>, Shuah Khan <shuah@kernel.org>,
linux-kselftest@vger.kernel.org,
Weiming Shi <bestswngs@gmail.com>,
Daniel Borkmann <daniel@iogearbox.net>,
Nikolay Aleksandrov <razor@blackwall.org>,
David Wei <dw@davidwei.uk>,
Stanislav Fomichev <sdf.kernel@gmail.com>,
Bobby Eshleman <bobbyeshleman@meta.com>
Subject: [PATCH net-next v4 0/7] net: devmem: support leased RX and TX queues on netdev unregister
Date: Fri, 20 Nov 2026 04:59:11 +0000 [thread overview]
Message-ID: <20261120045918.1239746-1-almasrymina@google.com> (raw)
Support leased RX and TX devmem bindings when either the virtual device
(e.g. netkit) or the physical backing net_device is unregistered:
1. Pass struct net_device **dev_ptr to net_devmem_lock_dev() so callers
can lock either binding->map_dev or binding->bind_dev.
2. Rename binding->dev and binding->vdev to binding->map_dev (physical
DMA device) and binding->bind_dev (device passed to bind-rx/bind-tx).
3. Introduce netif_lease_valid() and guard netif_mp_open_rxq(),
netif_mp_close_rxq(), and netif_rxq_cleanup_unlease() against an
unregistering physical device reached through a queue lease.
4. Fold net_devmem_lookup_dmabuf() into net_devmem_get_binding() so the
xarray lookup, validation, device check, and reference acquisition
run under a single rcu_read_lock() section.
5. Always populate bind_dev on RX and TX bindings and handle teardown
and netdev unregistration of both bind_dev and map_dev, waiting for
synchronize_net() after clearing NET_DEVMEM_BINDING_LIVE before
unmapping bindings whose map_dev is unregistering.
6. Resolve the physical backing map_dev in netdev_nl_get_dma_dev() for
leased RX bindings, lock map_dev around net_devmem_bind_dmabuf() on
bind-rx when map_dev differs from netdev, assert both devices are
locked, and check netif_lease_valid(map_dev).
7. Add kselftests in nl_netdev.py covering leased RX and TX devmem
bindings across both virtual-first and physical-first netdev
unregistration orders.
These issues were discovered during AI code review (Sashiko) of the
earlier non-leased devmem unregister series (and subsequent mailing list
review) and confirmed by code inspection (not observed in production or
reproduced on real hardware). Tested with the nl_netdev.py selftest on
netdevsim added in this series.
v4:
- Rebase onto latest net-next/main.
- Fold net_devmem_lookup_dmabuf() into net_devmem_get_binding() under a
single rcu_read_lock() (deferring net_devmem_dmabuf_binding_get()
until after all checks pass) and update net_devmem_dev_unregister() to
use a two-pass loop with a single synchronize_net() after clearing
NET_DEVMEM_BINDING_LIVE and before unmapping bindings with
map_dev == dev (Paolo).
- Document issue discovery, validation, and testing in commit messages
and the cover letter (netdev-bot+sinfo).
- v3: https://lore.kernel.org/netdev/20261115022607.348112-1-almasrymina@google.com/
v3:
- Rebase onto latest net-next/main.
- Add netif_lease_valid() helper and keep lease validity checks in the
lease-aware callers (netif_mp_open_rxq(), netif_mp_close_rxq(),
netif_rxq_cleanup_unlease(), and net_devmem_bind_dmabuf()) rather than
inside __netif_mp_open_rxq() / __netif_mp_close_rxq(), and document why
mp_params is cleared directly when !netif_lease_valid(dev) (Stanislav,
Sashiko).
- Guard the bound_rxqs loop in net_devmem_unbind_dmabuf() with
if (binding->bind_dev), keep the !netif_is_queue_leasee(dev) guard in
net_devmem_dev_unregister() in patch 4/6 until patch 5/6 resolves
binding->map_dev to the physical device so the series is bisectable at
every commit, and document the use-after-free in dma_unmap_sg_attrs() on
physical netdev teardown in patch 5/6 (Sashiko).
- In nl_netdev.py, verify dmabuf and lease on the physical queue after
binding, skip when queue_create returns EOPNOTSUPP, probe the physical
device on EOPNOTSUPP so a broken lease lookup is not masked as SKIP, and
clean up nk_ifname between loop iterations (Sashiko).
- v2: https://lore.kernel.org/netdev/20261108055809.2122934-1-almasrymina@google.com/
v2:
- Split patch 3/4 from v1 into 3 smaller commits (patches 3/6..5/6).
- Clear rxq->mp_params and return 0 in __netif_mp_close_rxq() whenever
dev is already unregistering, avoiding queue reconfiguration on an
already-shut-down map_dev during cascaded lease teardown (Sashiko).
- Drop unnecessary xa_erase() in net_devmem_unbind_dmabuf() and use a
local map_dev variable in net_devmem_get_binding() (Stanislav).
- Restore ASSERT_EXCLUSIVE_WRITER(binding->bind_dev) in
mp_dmabuf_devmem_uninstall(), drop redundant bind_dev->reg_state check
in net_devmem_bind_dmabuf(), and expand commit message rationale
(Dragos, Stanislav).
- v1: https://lore.kernel.org/netdev/20261103232659.1340283-1-almasrymina@google.com/
Mina Almasry (7):
net: devmem: pass net_device pointer to net_devmem_lock_dev()
net: devmem: rename binding dev and vdev to map_dev and bind_dev
net: guard memory provider RX queue open and close on unregister
net: devmem: fold net_devmem_lookup_dmabuf() into
net_devmem_get_binding()
net: devmem: always populate bind_dev and detach both devices on
unregister
net: devmem: resolve physical map_dev for leased RX bindings
selftests: net: add leased devmem RX and TX netdev unregister tests
net/core/dev.c | 2 +-
net/core/dev.h | 7 +
net/core/devmem.c | 178 ++++++++++++-----------
net/core/devmem.h | 31 ++--
net/core/netdev-genl.c | 59 +++++---
net/core/netdev_rx_queue.c | 31 +++-
tools/testing/selftests/net/nl_netdev.py | 159 +++++++++++++++++++-
7 files changed, 333 insertions(+), 134 deletions(-)
base-commit: 2208ab00f4f213ade91ab775e7f8c4dacdcc856e
--
2.56.1.750.gf70dc682cf-goog
next reply other threads:[~2026-11-20 5:49 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-20 4:59 Mina Almasry [this message]
2026-11-20 4:59 ` [PATCH net-next v4 1/7] net: devmem: pass net_device pointer to net_devmem_lock_dev() Mina Almasry
2026-11-20 4:59 ` [PATCH net-next v4 2/7] net: devmem: rename binding dev and vdev to map_dev and bind_dev Mina Almasry
2026-11-20 4:59 ` [PATCH net-next v4 3/7] net: guard memory provider RX queue open and close on unregister Mina Almasry
2026-11-20 4:59 ` [PATCH net-next v4 4/7] net: devmem: fold net_devmem_lookup_dmabuf() into net_devmem_get_binding() Mina Almasry
2026-11-20 4:59 ` [PATCH net-next v4 5/7] net: devmem: always populate bind_dev and detach both devices on unregister Mina Almasry
2026-11-20 4:59 ` [PATCH net-next v4 6/7] net: devmem: resolve physical map_dev for leased RX bindings Mina Almasry
2026-11-20 4:59 ` [PATCH net-next v4 7/7] selftests: net: add leased devmem RX and TX netdev unregister tests Mina Almasry
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=20261120045918.1239746-1-almasrymina@google.com \
--to=almasrymina@google.com \
--cc=bestswngs@gmail.com \
--cc=bobbyeshleman@meta.com \
--cc=daniel@iogearbox.net \
--cc=davem@davemloft.net \
--cc=dw@davidwei.uk \
--cc=edumazet@kernel.org \
--cc=horms@kernel.org \
--cc=kuba@kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=razor@blackwall.org \
--cc=sdf.kernel@gmail.com \
--cc=shuah@kernel.org \
/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