Netdev List
 help / color / mirror / Atom feed
From: Rishikesh Jethwani <rjethwani@purestorage.com>
To: netdev@vger.kernel.org
Cc: saeedm@nvidia.com, tariqt@nvidia.com, mbloch@nvidia.com,
	borisp@nvidia.com, john.fastabend@gmail.com, kuba@kernel.org,
	sd@queasysnail.net, davem@davemloft.net, pabeni@redhat.com,
	edumazet@google.com, leon@kernel.org,
	Rishikesh Jethwani <rjethwani@purestorage.com>
Subject: [PATCH net-next v14 0/9] tls: Add TLS 1.3 hardware offload support
Date: Fri, 15 May 2026 15:27:06 -0600	[thread overview]
Message-ID: <20260515212715.3151307-1-rjethwani@purestorage.com> (raw)

Hi all,

This series adds TLS 1.3 hardware offload support including KeyUpdate
(rekey) and a selftest for validation.

Patch 1: Reject TLS 1.3 offload in chcr_ktls and nfp drivers
These drivers only support TLS 1.2; add explicit version check.

Patch 2: mlx5e TLS 1.3 hardware offload
Add TLS 1.3 TX/RX offload on ConnectX-6 Dx and newer.
Handle 12-byte IV format and TLS_1_3 context type.

Patch 3: Core TLS 1.3 hardware offload support
Extend tls_device.c for TLS 1.3 record format (content type
appended before tag). Handle TLS 1.3 IV construction in fallback.

Patch 4: Split tls_set_sw_offload into init/finalize
Allows HW RX path to init SW context, attempt HW setup, then
finalize. Required for proper rekey error handling.

Patch 5: Prep helpers and refactors for HW offload KeyUpdate
No functional change. Hoist cipher_context/tls_crypto_context for
embedding in offload contexts. Factor tls_device_dev_add_tx() and
tls_device_commit_start_marker() for reuse by the rekey completion
path. Split tls_set_device_offload() into a dispatcher and
_initial() sibling. Move crypto_aead_setauthsize() into the !*aead
block so a fresh AEAD is correctly configured on RX HW rekey.

Patch 6: TX KeyUpdate support
tls_device_start_rekey() installs a temporary SW context with the
new key and redirects sendmsg. If no records are pending,
complete_rekey() runs inline; otherwise tls_tcp_clean_acked() sets
REKEY_READY once all old-key records are ACKed and the next sendmsg
completes the switch, flushing SW records and reinstalling HW at
the current write_seq. A KeyUpdate arriving during a pending rekey
re-keys the SW AEAD in place; if HW reinstall fails the socket
stays in SW mode (REKEY_FAILED). Adds TlsTxRekeyFallback and
TlsTxRekeyInProgress counters.

Patch 7: RX KeyUpdate support
tls_device_del_key_rx() is called from tls_check_pending_rekey()
when a KeyUpdate record is decoded. Old AEAD, IV and rec_seq are
retained on tls_offload_context_rx. tls_device_decrypted()
classifies records by old_nic_boundary: post-boundary records use
the new key; pre-boundary fully-encrypted records are decrypted by
SW AEAD; pre-boundary partially-decrypted records are reencrypted
with the old key for SW AEAD to decrypt with the new key. Mixed
records retry once with toggled decrypted flags (old_key_reencrypted
gate). The new key's tls_dev_add is deferred until copied_seq
crosses old_nic_boundary. Adds TlsRxRekeyFallback and
TlsRxRekeyInProgress counters.

Patch 8: Tracepoints for RX KeyUpdate path
Three trace events for the RX rekey state machine:
tls_device_rekey_start (inflight flag), tls_device_rekey_reencrypt
(old-key undo, retry flag), tls_device_rekey_done (old_aead_recv
freed, deferred dev_add issued).

Patch 9: Selftest for hardware offload
Python wrapper + C binary using NetDrvEpEnv framework.
Tests TLS 1.2/1.3, AES-GCM-128/256, rekey with various buffer
sizes, and burst variants stressing TX rekey (temporary SW phase,
HW reinstall) and RX rekey (boundary tracking, old-key
reencryption, deferred dev_add). Verifies RekeyOk, RekeyReceived,
RekeyFallback, RekeyInProgress, and DecryptError stat counters.

Rishikesh

Changes in v14:
  - Split the monolithic rekey patch into four patches (5-8) for
    easier review: prep/refactors, TX KeyUpdate, RX KeyUpdate,
    tracepoints.
  - Renamed TlsTxRekeyHwFail/TlsRxRekeyHwFail to
    TlsTxRekeyFallback/TlsRxRekeyFallback to better reflect that
    the counter tracks SW fallback, not just HW failure.
  - Added TlsTxRekeyInProgress/TlsRxRekeyInProgress counters to
    expose in-flight rekey state.
  - Selftest: updated stat counter names to match above renames.

Rishikesh Jethwani (9):
  net: tls: reject TLS 1.3 offload in chcr_ktls and nfp drivers
  net/mlx5e: add TLS 1.3 hardware offload support
  tls: add TLS 1.3 hardware offload support
  tls: split tls_set_sw_offload into init and finalize stages
  tls: prep helpers and refactors for HW offload KeyUpdate
  tls: device: add TX KeyUpdate support
  tls: device: add RX KeyUpdate support
  tls: device: add tracepoints for RX KeyUpdate path
  selftests: net: add TLS hardware offload test

 MAINTAINERS                                   |   2 +
 .../chelsio/inline_crypto/ch_ktls/chcr_ktls.c |   3 +
 .../mellanox/mlx5/core/en_accel/ktls.h        |   8 +-
 .../mellanox/mlx5/core/en_accel/ktls_txrx.c   |  14 +-
 .../net/ethernet/netronome/nfp/crypto/tls.c   |   3 +
 include/net/tls.h                             |  90 +-
 include/uapi/linux/snmp.h                     |   4 +
 net/tls/tls.h                                 |  31 +-
 net/tls/tls_device.c                          | 838 +++++++++++++--
 net/tls/tls_device_fallback.c                 |  82 +-
 net/tls/tls_main.c                            |  29 +-
 net/tls/tls_proc.c                            |   4 +
 net/tls/tls_sw.c                              | 165 ++-
 net/tls/trace.h                               |  79 ++
 .../selftests/drivers/net/hw/.gitignore       |   1 +
 .../testing/selftests/drivers/net/hw/Makefile |   2 +
 .../selftests/drivers/net/hw/tls_hw_offload.c | 971 ++++++++++++++++++
 .../drivers/net/hw/tls_hw_offload.py          | 257 +++++
 18 files changed, 2395 insertions(+), 188 deletions(-)
 create mode 100644 tools/testing/selftests/drivers/net/hw/tls_hw_offload.c
 create mode 100755 tools/testing/selftests/drivers/net/hw/tls_hw_offload.py

-- 
2.25.1


             reply	other threads:[~2026-05-15 21:28 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-15 21:27 Rishikesh Jethwani [this message]
2026-05-15 21:27 ` [PATCH v14 1/9] net: tls: reject TLS 1.3 offload in chcr_ktls and nfp drivers Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 2/9] net/mlx5e: add TLS 1.3 hardware offload support Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 3/9] tls: " Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 4/9] tls: split tls_set_sw_offload into init and finalize stages Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 5/9] tls: prep helpers and refactors for HW offload KeyUpdate Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 6/9] tls: device: add TX KeyUpdate support Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 7/9] tls: device: add RX " Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 8/9] tls: device: add tracepoints for RX KeyUpdate path Rishikesh Jethwani
2026-05-15 21:27 ` [PATCH v14 9/9] selftests: net: add TLS hardware offload test Rishikesh Jethwani

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=20260515212715.3151307-1-rjethwani@purestorage.com \
    --to=rjethwani@purestorage.com \
    --cc=borisp@nvidia.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=john.fastabend@gmail.com \
    --cc=kuba@kernel.org \
    --cc=leon@kernel.org \
    --cc=mbloch@nvidia.com \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=saeedm@nvidia.com \
    --cc=sd@queasysnail.net \
    --cc=tariqt@nvidia.com \
    /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