[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20260204185220.0ef1aed9@kernel.org>
Date: Wed, 4 Feb 2026 18:52:20 -0800
From: Jakub Kicinski <kuba@...nel.org>
To: Rishikesh Jethwani <rjethwani@...estorage.com>
Cc: netdev@...r.kernel.org, saeedm@...dia.com, tariqt@...dia.com,
mbloch@...dia.com, borisp@...dia.com, john.fastabend@...il.com,
sd@...asysnail.net, davem@...emloft.net, pabeni@...hat.com,
edumazet@...gle.com, leon@...nel.org
Subject: Re: [PATCH v6 4/4] selftests: tls: add two-node hardware offload
test
On Tue, 3 Feb 2026 11:48:35 -0700 Rishikesh Jethwani wrote:
> Add tls_hw_offload, a test program for validating TLS hardware offload
> functionality across two physical nodes. Unlike existing selftests that
> use loopback or veth pairs, this test requires separate server and
> client machines to trigger actual NIC hardware offload.
>
> Features:
> - Server/client mode for two-node testing
> - TLS 1.2 and TLS 1.3 support
> - AES-GCM-128 and AES-GCM-256 cipher selection
> - TLS 1.3 KeyUpdate (rekey) testing with configurable count
> - Verification of /proc/net/tls_stat counters
> - Echo protocol with data integrity verification
> - Configurable buffer sizes (fixed or random)
>
> The test validates that TLS connections use hardware offload by checking
> TlsTxDevice/TlsRxDevice counters, and verifies rekey operations through
> TlsTxRekeyOk/TlsRxRekeyOk counters. It accepts both hardware and software
> TLS paths, reporting which was used.
I appreciate the LLM tokens spent on this. Please get this rewritten
to use our existing driver test setup instead of creating your own.
See tools/testing/selftests/drivers/net/README.rst and the existing
tests under tools/testing/selftests/drivers/net/hw/
Please switch to posting as RFC until we get a review from Sabrina,
the merge window is soon, anyways..
Powered by blists - more mailing lists