This series improves the CPU cost of RX token management by adding a socket option that configures the socket to avoid the xarray allocator and instead use an niov array and a uref field in niov.
Improvement is ~13% cpu util per RX user thread.
Using kperf, the following results were observed:
Before: Average RX worker idle %: 13.13, flows 4, test runs 11 After: Average RX worker idle %: 26.32, flows 4, test runs 11
Two other approaches were tested, but with no improvement. Namely, 1) using a hashmap for tokens and 2) keeping an xarray of atomic counters but using RCU so that the hotpath could be mostly lockless. Neither of these approaches proved better than the simple array in terms of CPU.
The sockopt SO_DEVMEM_AUTORELEASE is added to toggle the optimization. It defaults to 0 (i.e., optimization on).
Note that prior revs reported only a 5% gain. This lower gain was measured with cpu frequency boosting (unknowingly) disabled. A consistent ~13% is measured for both kperf and nccl workloads with cpu frequency boosting on.
To: David S. Miller davem@davemloft.net To: Eric Dumazet edumazet@google.com To: Jakub Kicinski kuba@kernel.org To: Paolo Abeni pabeni@redhat.com To: Simon Horman horms@kernel.org To: Kuniyuki Iwashima kuniyu@google.com To: Willem de Bruijn willemb@google.com To: Neal Cardwell ncardwell@google.com To: David Ahern dsahern@kernel.org To: Mina Almasry almasrymina@google.com To: Arnd Bergmann arnd@arndb.de To: Jonathan Corbet corbet@lwn.net To: Andrew Lunn andrew+netdev@lunn.ch To: Shuah Khan shuah@kernel.org Cc: Stanislav Fomichev sdf@fomichev.me Cc: netdev@vger.kernel.org Cc: linux-kernel@vger.kernel.org Cc: linux-arch@vger.kernel.org Cc: linux-doc@vger.kernel.org Cc: linux-kselftest@vger.kernel.org Signed-off-by: Bobby Eshleman bobbyeshleman@meta.com
Changes in v6: - renamed 'net: devmem: use niov array for token management' to refer to optionality of new config - added documentation and tests - make autorelease flag per-socket sockopt instead of binding field / sysctl - many per-patch changes (see Changes sections per-patch) - Link to v5: https://lore.kernel.org/r/20251023-scratch-bobbyeshleman-devmem-tcp-token-up...
Changes in v5: - add sysctl to opt-out of performance benefit, back to old token release - Link to v4: https://lore.kernel.org/all/20250926-scratch-bobbyeshleman-devmem-tcp-token-...
Changes in v4: - rebase to net-next - Link to v3: https://lore.kernel.org/r/20250926-scratch-bobbyeshleman-devmem-tcp-token-up...
Changes in v3: - make urefs per-binding instead of per-socket, reducing memory footprint - fallback to cleaning up references in dmabuf unbind if socket leaked tokens - drop ethtool patch - Link to v2: https://lore.kernel.org/r/20250911-scratch-bobbyeshleman-devmem-tcp-token-up...
Changes in v2: - net: ethtool: prevent user from breaking devmem single-binding rule (Mina) - pre-assign niovs in binding->vec for RX case (Mina) - remove WARNs on invalid user input (Mina) - remove extraneous binding ref get (Mina) - remove WARN for changed binding (Mina) - always use GFP_ZERO for binding->vec (Mina) - fix length of alloc for urefs - use atomic_set(, 0) to initialize sk_user_frags.urefs - Link to v1: https://lore.kernel.org/r/20250902-scratch-bobbyeshleman-devmem-tcp-token-up...
--- Bobby Eshleman (6): net: devmem: rename tx_vec to vec in dmabuf binding net: devmem: refactor sock_devmem_dontneed for autorelease split net: devmem: prepare for autorelease rx token management net: devmem: add SO_DEVMEM_AUTORELEASE for autorelease control net: devmem: document SO_DEVMEM_AUTORELEASE socket option net: devmem: add tests for SO_DEVMEM_AUTORELEASE socket option
Documentation/networking/devmem.rst | 70 +++++++++- include/net/netmem.h | 1 + include/net/sock.h | 13 +- include/uapi/asm-generic/socket.h | 2 + net/core/devmem.c | 54 +++++--- net/core/devmem.h | 4 +- net/core/sock.c | 152 ++++++++++++++++++---- net/ipv4/tcp.c | 69 ++++++++-- net/ipv4/tcp_ipv4.c | 11 +- net/ipv4/tcp_minisocks.c | 5 +- tools/include/uapi/asm-generic/socket.h | 2 + tools/testing/selftests/drivers/net/hw/devmem.py | 115 +++++++++++++++- tools/testing/selftests/drivers/net/hw/ncdevmem.c | 20 ++- 13 files changed, 453 insertions(+), 65 deletions(-) --- base-commit: 255d75ef029f33f75fcf5015052b7302486f7ad2 change-id: 20250829-scratch-bobbyeshleman-devmem-tcp-token-upstream-292be174d503
Best regards,