From: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
To: dev@dpdk.org
Cc: ferruh.yigit@amd.com, kai.ji@intel.com, brian.dooley@intel.com,
Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
Subject: [PATCH v2 2/3] crypto/qat: use process safe crc api
Date: Tue, 1 Oct 2024 19:11:49 +0100 [thread overview]
Message-ID: <20241001181150.43506-3-arkadiuszx.kusztal@intel.com> (raw)
In-Reply-To: <20241001181150.43506-1-arkadiuszx.kusztal@intel.com>
This patch replaces thread-unsafe CRC functions
with the safe ones.
Signed-off-by: Arkadiusz Kusztal <arkadiuszx.kusztal@intel.com>
---
doc/guides/cryptodevs/qat.rst | 6 ++++++
drivers/crypto/qat/qat_sym.h | 6 ++----
drivers/crypto/qat/qat_sym_session.c | 3 +++
drivers/crypto/qat/qat_sym_session.h | 2 ++
4 files changed, 13 insertions(+), 4 deletions(-)
diff --git a/doc/guides/cryptodevs/qat.rst b/doc/guides/cryptodevs/qat.rst
index 68d792e4cc..d98195300b 100644
--- a/doc/guides/cryptodevs/qat.rst
+++ b/doc/guides/cryptodevs/qat.rst
@@ -148,6 +148,12 @@ Limitations
DOCSIS security protocol.
* Multi-segment buffers are not supported for combined Crypto-CRC DOCSIS
security protocol.
+* Max SIMD bitwidth (--force-max-simd-bitwidth) in multi-process may be ignored
+ by QAT in certain scenarios. In use cases where CRC computation is needed and
+ the device does not support it internally, the CRC context is created by the session;
+ therefore, the SIMD config used in the process that created the session will be persisant
+ across all processes. It is an undefined behavior when different process binaries are
+ compiled with different SIMD capabilities when using software CRC.
Extra notes on KASUMI F9
~~~~~~~~~~~~~~~~~~~~~~~~
diff --git a/drivers/crypto/qat/qat_sym.h b/drivers/crypto/qat/qat_sym.h
index eedf5de755..d0fda48e01 100644
--- a/drivers/crypto/qat/qat_sym.h
+++ b/drivers/crypto/qat/qat_sym.h
@@ -267,8 +267,7 @@ qat_crc_verify(struct qat_sym_session *ctx, struct rte_crypto_op *op)
crc_data = rte_pktmbuf_mtod_offset(sym_op->m_src, uint8_t *,
crc_data_ofs);
- crc = rte_net_crc_calc(crc_data, crc_data_len,
- RTE_NET_CRC32_ETH);
+ crc = rte_net_crc(&ctx->crc, crc_data, crc_data_len);
if (crc != *(uint32_t *)(crc_data + crc_data_len))
op->status = RTE_CRYPTO_OP_STATUS_AUTH_FAILED;
@@ -291,8 +290,7 @@ qat_crc_generate(struct qat_sym_session *ctx,
crc_data = rte_pktmbuf_mtod_offset(sym_op->m_src, uint8_t *,
sym_op->auth.data.offset);
crc = (uint32_t *)(crc_data + crc_data_len);
- *crc = rte_net_crc_calc(crc_data, crc_data_len,
- RTE_NET_CRC32_ETH);
+ *crc = rte_net_crc(&ctx->crc, crc_data, crc_data_len);
}
}
diff --git a/drivers/crypto/qat/qat_sym_session.c b/drivers/crypto/qat/qat_sym_session.c
index eb267db424..4f281de9c5 100644
--- a/drivers/crypto/qat/qat_sym_session.c
+++ b/drivers/crypto/qat/qat_sym_session.c
@@ -3176,6 +3176,9 @@ qat_sec_session_set_docsis_parameters(struct rte_cryptodev *dev,
ret = qat_sym_session_configure_crc(dev, xform, session);
if (ret < 0)
return ret;
+ } else {
+ /* Initialize crc algorithm */
+ session->crc = rte_net_crc_set(RTE_NET_CRC_AVX512, RTE_NET_CRC32_ETH);
}
qat_sym_session_finalize(session);
diff --git a/drivers/crypto/qat/qat_sym_session.h b/drivers/crypto/qat/qat_sym_session.h
index f2634774ec..fb84d0f494 100644
--- a/drivers/crypto/qat/qat_sym_session.h
+++ b/drivers/crypto/qat/qat_sym_session.h
@@ -7,6 +7,7 @@
#include <rte_crypto.h>
#include <cryptodev_pmd.h>
#include <rte_security.h>
+#include <rte_net_crc.h>
#include "qat_common.h"
#include "icp_qat_hw.h"
@@ -151,6 +152,7 @@ struct qat_sym_session {
uint32_t slice_types;
enum qat_sym_proto_flag qat_proto_flag;
qat_sym_build_request_t build_request[2];
+ struct rte_net_crc crc;
#ifndef RTE_QAT_OPENSSL
IMB_MGR *mb_mgr;
alignas(16) uint64_t expkey[4 * 15];
--
2.13.6
next prev parent reply other threads:[~2024-10-01 19:27 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-01 18:11 [PATCH v2 0/3] net: add thread-safe " Arkadiusz Kusztal
2024-10-01 18:11 ` [PATCH v2 1/3] " Arkadiusz Kusztal
2024-10-01 21:44 ` Stephen Hemminger
2024-10-01 18:11 ` Arkadiusz Kusztal [this message]
2024-10-01 18:11 ` [PATCH v2 3/3] test/crc: replace thread-unsafe api functions Arkadiusz Kusztal
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=20241001181150.43506-3-arkadiuszx.kusztal@intel.com \
--to=arkadiuszx.kusztal@intel.com \
--cc=brian.dooley@intel.com \
--cc=dev@dpdk.org \
--cc=ferruh.yigit@amd.com \
--cc=kai.ji@intel.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;
as well as URLs for NNTP newsgroup(s).