app: test: measure libipsec performance
PATCH 1/1 app: test: measure libipsec performance
test app to measures the performance of libipsec api rte_ipsec_pkt_crypto and rte_ipsec_pkt_process.
--- app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 637 +++++++++++++++++++++++++++++++++++++ 3 files changed, 640 insertions(+), 1 deletion(-) create mode 100644 app/test/test_ipsec_perf.c
app/test/Makefile +1 -1
@@ -231,7 +231,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BPF) += test_bpf.c SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsec
app/test/meson.build +2 -0
@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',
@@ -225,6 +226,7 @@ fast_test_names = [ 'fbarray_autotest', 'hash_readwrite_func_autotest', 'ipsec_autotest', + 'ipsec_perf_autotest', 'kni_autotest', 'kvargs_autotest', 'member_autotest',
app/test/test_ipsec_perf.c new file +637 -0
637 lines not on this page · raw message
Signature · 1 line
2.17.1
PATCH v2 1/1 app: test: measure libipsec performance
test app to measures the performance of libipsec api rte_ipsec_pkt_crypto and rte_ipsec_pkt_process.
--- app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 637 +++++++++++++++++++++++++++++++++++++ 3 files changed, 640 insertions(+), 1 deletion(-) create mode 100644 app/test/test_ipsec_perf.c
app/test/Makefile +1 -1
@@ -231,7 +231,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BPF) += test_bpf.c SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsec
app/test/meson.build +2 -0
@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',
@@ -225,6 +226,7 @@ fast_test_names = [ 'fbarray_autotest', 'hash_readwrite_func_autotest', 'ipsec_autotest', + 'ipsec_perf_autotest', 'kni_autotest', 'kvargs_autotest', 'member_autotest',
app/test/test_ipsec_perf.c new file +637 -0
637 lines not on this page · raw message
Signature · 1 line
2.17.1
Quoted 5 lines · Savinay Dharmappa
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.Signed-off-by Savinay Dharmappa <[email protected]>
Missing MAINTAINERS update.
Quoted 33 lines
--- app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 637 +++++++++++++++++++++++++++++++++++++ 3 files changed, 640 insertions(+), 1 deletion(-) create mode 100644 app/test/test_ipsec_perf.capp/test/Makefile +1
-1@@ -231,7 +231,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BPF) += test_bpf.c SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsecapp/test/meson.build +1
-0@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',
app/test/meson.build fast_test_names = [ · line 232
Quoted 7 lines
app/test/meson.build +1
-0@@ -225,6 +226,7 @@ fast_test_names = [ 'fbarray_autotest', 'hash_readwrite_func_autotest', 'ipsec_autotest', + 'ipsec_perf_autotest', 'kni_autotest', 'kvargs_autotest', 'member_autotest',
This test ends with a timeout in Travis.
Perf tests should be in the appropriate list.
https://travis-ci.com/ovsrobot/dpdk/jobs/292924971
Signature · 1 line
David Marchand
Hi Savinay,
Quoted 74 lines · Savinay Dharmappa
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.Signed-off-by Savinay Dharmappa <[email protected]>--- app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 637 +++++++++++++++++++++++++++++++++++++ 3 files changed, 640 insertions(+), 1 deletion(-)create mode 100644
app/test/test_ipsec_perf.capp/test/Makefile +1
-1@@ -231,7 +231,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BPF) += test_bpf.c SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsecapp/test/meson.build +2
-0@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',@@ -225,6 +226,7 @@ fast_test_names = [ 'fbarray_autotest', 'hash_readwrite_func_autotest', 'ipsec_autotest', + 'ipsec_perf_autotest', 'kni_autotest', 'kvargs_autotest', 'member_autotest',app/test/test_ipsec_perf.c new file +24
-0@@ -0,0 +1,637 @@ +/* SPDX-License-Identifier: BSD-3-Clause +* Copyright(c) 2010-2014 Intel Corporation +*/ + +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <stdint.h> +#include <unistd.h> + +#include <rte_cycles.h> +#include <rte_ether.h> +#include <rte_ip.h> +#include <rte_byteorder.h> +#include <rte_malloc.h> +#include <rte_ring.h> +#include <rte_memory.h> +#include <rte_common.h> +#include <rte_hexdump.h> +#include <rte_mbuf.h> +#include <rte_malloc.h> +#include <rte_memcpy.h> +#include <rte_cycles.h> +#include <rte_bus_vdev.h>
app/test/test_ipsec_perf.c line 34
Quoted 10 lines
app/test/test_ipsec_perf.c new file +10
-0+#include <rte_ip.h> + +#include <rte_crypto.h> +#include <rte_cryptodev.h> +#include <rte_cryptodev_pmd.h> +#include <rte_lcore.h> +#include <rte_ipsec.h> +#include <rte_random.h> +#include <rte_esp.h> +#include <rte_security_driver.h>
Please remove unnecessary includes
Quoted 603 lines
app/test/test_ipsec_perf.c new file +603
-0+ +#include "test.h" +#include "test_cryptodev.h" + +#define RING_SIZE 4096 +#define BURST_SIZE 64 +#define NUM_MBUF 4095 +#define DEFAULT_SPI 7 + +struct ipsec_test_cfg { +uint32_t replay_win_sz; +uint32_t esn; +uint64_t flags; +enum rte_crypto_sym_xform_type type; +}; + +struct rte_mempool *mbuf_pool, *cop_pool; + +struct stats_counter { +uint64_t nb_prepare_call; +uint64_t nb_prepare_pkt; +uint64_t nb_process_call; +uint64_t nb_process_pkt; +uint64_t prepare_ticks_elapsed; +uint64_t process_ticks_elapsed; +}; + +struct ipsec_sa { +struct rte_ipsec_session ss[2 ]; +struct rte_ipsec_sa_prm sa_prm; +struct rte_security_ipsec_xform ipsec_xform; +struct rte_crypto_sym_xform cipher_xform; +struct rte_crypto_sym_xform auth_xform; +struct rte_crypto_sym_xform aead_xform; +struct rte_crypto_sym_xform *crypto_xforms; +struct rte_crypto_op *cop[BURST_SIZE ]; +enum rte_crypto_sym_xform_type type; +struct stats_counter cnt; +uint32_t replay_win_sz; +uint32_t sa_flags; +}; + +static const struct ipsec_test_cfg test_cfg[] = { + {0 ,0 ,0 ,RTE_CRYPTO_SYM_XFORM_AEAD }, + {0 ,0 ,0 ,RTE_CRYPTO_SYM_XFORM_CIPHER }, + {128 ,1 ,0 ,RTE_CRYPTO_SYM_XFORM_AEAD }, + {128 ,1 ,0 ,RTE_CRYPTO_SYM_XFORM_CIPHER }, + +}; + +static struct rte_ipv4_hdr ipv4_outer = { + .version_ihl =IPVERSION <<4 | +sizeof (ipv4_outer) /RTE_IPV4_IHL_MULTIPLIER , + .time_to_live =IPDEFTTL , + .next_proto_id =IPPROTO_ESP , + .src_addr =RTE_IPV4 (192 ,168 ,1 ,100 ), + .dst_addr =RTE_IPV4 (192 ,168 ,2 ,100 ), +}; + +static struct rte_ring *ring_inb_prepare; +static struct rte_ring *ring_inb_process; +static struct rte_ring *ring_outb_prepare; +static struct rte_ring *ring_outb_process; + +struct supported_cipher_algo { +const char *keyword; +enum rte_crypto_cipher_algorithm algo; +uint16_t iv_len; +uint16_t block_size; +uint16_t key_len; +}; + +struct supported_auth_algo { +const char *keyword; +enum rte_crypto_auth_algorithm algo; +uint16_t digest_len; +uint16_t key_len; +uint8_t key_not_req; +}; + +struct supported_aead_algo { +const char *keyword; +enum rte_crypto_aead_algorithm algo; +uint16_t iv_len; +uint16_t block_size; +uint16_t digest_len; +uint16_t key_len; +uint8_t aad_len; +}; + +const struct supported_cipher_algo cipher_algo[] = { + { + .keyword ="aes-128-cbc" , + .algo =RTE_CRYPTO_CIPHER_AES_CBC , + .iv_len =16 , + .block_size =16 , + .key_len =16 + } +}; + +const struct supported_auth_algo auth_algo[] = { + { + .keyword ="sha1-hmac" , + .algo =RTE_CRYPTO_AUTH_SHA1_HMAC , + .digest_len =12 , + .key_len =20 + } +}; + +const struct supported_aead_algo aead_algo[] = { + { + .keyword ="aes-128-gcm" , + .algo =RTE_CRYPTO_AEAD_AES_GCM , + .iv_len =8 , + .block_size =4 , + .key_len =20 , + .digest_len =16 , + .aad_len =8 , + } +}; + +static struct rte_mbuf *generate_mbuf_data (struct rte_mempool *mpool) +{ +struct rte_mbuf *mbuf =rte_pktmbuf_alloc (mpool); + +if (mbuf) { + mbuf->data_len =64 ; + mbuf->pkt_len =64 ; + } + +return mbuf; +} + +static int +fill_ipsec_param (struct ipsec_sa *sa) +{ +struct rte_ipsec_sa_prm *prm = &sa->sa_prm; + +memset (prm,0 ,sizeof (*prm)); + + prm->flags = sa->sa_flags; + +/* setup ipsec xform */ + prm->ipsec_xform = sa->ipsec_xform; + prm->ipsec_xform.salt = (uint32_t )rte_rand (); + prm->ipsec_xform.replay_win_sz = sa->replay_win_sz; + +/* setup tunnel related fields */ + prm->tun.hdr_len =sizeof (ipv4_outer); + prm->tun.next_proto =IPPROTO_IPIP ; + prm->tun.hdr = &ipv4_outer; + +if (sa->type ==RTE_CRYPTO_SYM_XFORM_AEAD ) { + sa->aead_xform.type = sa->type; + sa->aead_xform.aead.algo = aead_algo->algo; + sa->aead_xform.next =NULL ; + sa->aead_xform.aead.digest_length = aead_algo->digest_len; + sa->aead_xform.aead.iv.offset =IV_OFFSET ; + sa->aead_xform.aead.iv.length =12 ; + +if (sa->ipsec_xform.direction == +RTE_SECURITY_IPSEC_SA_DIR_INGRESS ) { + sa->aead_xform.aead.op =RTE_CRYPTO_AEAD_OP_DECRYPT ; + }else { + sa->aead_xform.aead.op =RTE_CRYPTO_AEAD_OP_ENCRYPT ; + } + + sa->crypto_xforms = &sa->aead_xform; + }else { + sa->cipher_xform.type =RTE_CRYPTO_SYM_XFORM_CIPHER ; + sa->cipher_xform.cipher.algo = cipher_algo->algo; + sa->cipher_xform.cipher.iv.offset =IV_OFFSET ; + sa->cipher_xform.cipher.iv.length =12 ; + sa->auth_xform.type =RTE_CRYPTO_SYM_XFORM_AUTH ; + sa->auth_xform.auth.algo = auth_algo->algo; + sa->auth_xform.auth.digest_length = auth_algo->digest_len; + + +if (sa->ipsec_xform.direction == +RTE_SECURITY_IPSEC_SA_DIR_INGRESS ) { + sa->cipher_xform.cipher.op = +RTE_CRYPTO_CIPHER_OP_DECRYPT ; + sa->auth_xform.auth.op =RTE_CRYPTO_AUTH_OP_VERIFY ; + sa->cipher_xform.next =NULL ; + sa->auth_xform.next = &sa->cipher_xform; + sa->crypto_xforms = &sa->auth_xform; + }else { + sa->cipher_xform.cipher.op = +RTE_CRYPTO_CIPHER_OP_ENCRYPT ; + sa->auth_xform.auth.op =RTE_CRYPTO_AUTH_OP_GENERATE ; + sa->auth_xform.next =NULL ; + sa->cipher_xform.next = &sa->auth_xform; + sa->crypto_xforms = &sa->cipher_xform; + } + } + + prm->crypto_xform = sa->crypto_xforms; + +return TEST_SUCCESS ; +} + +static int +create_sa (enum rte_security_session_action_type action_type, +struct ipsec_sa *sa) +{ +static struct rte_cryptodev_sym_session dummy_ses; +size_t sz; +int rc; + +memset (&sa->ss[0 ],0 ,sizeof (sa->ss[0 ])); + + rc =fill_ipsec_param (sa); +if (rc !=0 ) { +printf ("failed to fill ipsec param\n" ); +return TEST_FAILED ; + } + +/* create rte_ipsec_sa*/ + sz =rte_ipsec_sa_size (&sa->sa_prm); +TEST_ASSERT (sz >0 ,"rte_ipsec_sa_size() failed\n" ); + + sa->ss[0 ].sa =rte_zmalloc (NULL , sz,RTE_CACHE_LINE_SIZE ); +TEST_ASSERT_NOT_NULL (sa->ss[0 ].sa, +"failed to allocate memory for rte_ipsec_sa\n" ); + + sa->ss[0 ].type = action_type; + sa->ss[0 ].crypto.ses = &dummy_ses; + + rc =rte_ipsec_sa_init (sa->ss[0 ].sa, &sa->sa_prm, sz); + rc = (rc >0 && (uint32_t )rc <= sz) ?0 : -EINVAL ; + +if (rc ==0 ) + rc =rte_ipsec_session_prepare (&sa->ss[0 ]); +else +return TEST_FAILED ; + +return TEST_SUCCESS ; +} + +static int +packet_prepare (struct rte_mbuf **buf,struct ipsec_sa *sa, +uint16_t num_pkts) +{ +uint64_t time_stamp; +uint16_t k =0 ; + +for (int i =0 ; i < num_pkts; i++) { + + sa->cop[i] =rte_crypto_op_alloc (cop_pool, +RTE_CRYPTO_OP_TYPE_SYMMETRIC ); + +if (sa->cop[i] ==NULL ) { + +RTE_LOG (ERR ,USER1 , +"Failed to allocate symmetric crypto op\n" ); + +return k; + } + } + +/* call crypto prepare */ + time_stamp =rte_rdtsc_precise (); + + k =rte_ipsec_pkt_crypto_prepare (&sa->ss[0 ], buf, + sa->cop, num_pkts); + + time_stamp =rte_rdtsc_precise () - time_stamp; + +if (k != num_pkts) { +RTE_LOG (ERR ,USER1 ,"rte_ipsec_pkt_crypto_prepare fail\n" ); +return k; + } + + sa->cnt.prepare_ticks_elapsed += time_stamp; + sa->cnt.nb_prepare_call++; + sa->cnt.nb_prepare_pkt += k; + +for (int i =0 ; i < num_pkts; i++) +rte_crypto_op_free (sa->cop[i]); + +return k; +} + +static int +packet_process (struct rte_mbuf **buf,struct ipsec_sa *sa, +uint16_t num_pkts) +{ +uint64_t time_stamp; +uint16_t k =0 ; + + time_stamp =rte_rdtsc_precise (); + +/* call crypto prepare */ + k =rte_ipsec_pkt_process (&sa->ss[0 ], buf, num_pkts); + + time_stamp =rte_rdtsc_precise () - time_stamp; + +if (k != num_pkts) { +RTE_LOG (ERR ,USER1 ,"rte_ipsec_pkt_process fail\n" ); +return k; + } + + sa->cnt.process_ticks_elapsed += time_stamp; + sa->cnt.nb_process_call++; + sa->cnt.nb_process_pkt += k; + +return k; +} + +static int +create_traffic (struct ipsec_sa *sa,struct rte_ring *deq_ring, +struct rte_ring *enq_ring,struct rte_ring *ring) +{ +struct rte_mbuf *mbuf[BURST_SIZE ]; +uint16_t num_pkts, n; + +while (rte_ring_empty (deq_ring) ==0 ) { + + num_pkts =rte_ring_sc_dequeue_burst (deq_ring, (void **)mbuf, +RTE_DIM (mbuf),NULL ); + +if (num_pkts ==0 ) +return TEST_FAILED ; + + n =packet_prepare (mbuf, sa, num_pkts); +if (n != num_pkts) +return TEST_FAILED ; + + num_pkts =rte_ring_sp_enqueue_burst (enq_ring, (void **)mbuf, + num_pkts,NULL ); +if (num_pkts ==0 ) +return TEST_FAILED ; + } + + deq_ring = enq_ring; + enq_ring = ring; + +while (rte_ring_empty (deq_ring) ==0 ) { + + num_pkts =rte_ring_sc_dequeue_burst (deq_ring, (void **)mbuf, +RTE_DIM (mbuf),NULL ); +if (num_pkts ==0 ) +return TEST_FAILED ; + + n =packet_process (mbuf, sa, num_pkts); +if (n != num_pkts) +return TEST_FAILED ; + + num_pkts =rte_ring_sp_enqueue_burst (enq_ring, (void **)mbuf, + num_pkts,NULL ); +if (num_pkts ==0 ) +return TEST_FAILED ; + } + +return TEST_SUCCESS ; +} + +static void +fill_ipsec_sa_out (const struct ipsec_test_cfg *test_cfg, +struct ipsec_sa *sa) +{ + sa->ipsec_xform.spi =DEFAULT_SPI ; + sa->ipsec_xform.direction =RTE_SECURITY_IPSEC_SA_DIR_EGRESS ; + sa->ipsec_xform.proto =RTE_SECURITY_IPSEC_SA_PROTO_ESP ; + sa->ipsec_xform.mode =RTE_SECURITY_IPSEC_SA_MODE_TUNNEL ; + sa->ipsec_xform.tunnel.type =RTE_SECURITY_IPSEC_TUNNEL_IPV4 ; + sa->ipsec_xform.options.esn = test_cfg->esn; + sa->type = test_cfg->type; + sa->replay_win_sz = test_cfg->replay_win_sz; + sa->sa_flags = test_cfg->flags; + sa->cnt.nb_prepare_call =0 ; + sa->cnt.nb_prepare_pkt =0 ; + sa->cnt.nb_process_call =0 ; + sa->cnt.nb_process_pkt =0 ; + sa->cnt.process_ticks_elapsed =0 ; + sa->cnt.prepare_ticks_elapsed =0 ; + +} + +static void +fill_ipsec_sa_in (const struct ipsec_test_cfg *test_cfg, +struct ipsec_sa *sa) +{ + sa->ipsec_xform.spi =DEFAULT_SPI ; + sa->ipsec_xform.direction =RTE_SECURITY_IPSEC_SA_DIR_INGRESS ; + sa->ipsec_xform.proto =RTE_SECURITY_IPSEC_SA_PROTO_ESP ; + sa->ipsec_xform.mode =RTE_SECURITY_IPSEC_SA_MODE_TUNNEL ; + sa->ipsec_xform.tunnel.type =RTE_SECURITY_IPSEC_TUNNEL_IPV4 ; + sa->ipsec_xform.options.esn = test_cfg->esn; + sa->type = test_cfg->type; + sa->replay_win_sz = test_cfg->replay_win_sz; + sa->sa_flags = test_cfg->flags; + sa->cnt.nb_prepare_call =0 ; + sa->cnt.nb_prepare_pkt =0 ; + sa->cnt.nb_process_call =0 ; + sa->cnt.nb_process_pkt =0 ; + sa->cnt.process_ticks_elapsed =0 ; + sa->cnt.prepare_ticks_elapsed =0 ; +} + +static int +init_sa_session (const struct ipsec_test_cfg *test_cfg, +struct ipsec_sa *sa_out,struct ipsec_sa *sa_in) +{ + +int rc; + +fill_ipsec_sa_in (test_cfg, sa_in); +fill_ipsec_sa_out (test_cfg, sa_out); + +/* create rte_ipsec_sa*/ + rc =create_sa (RTE_SECURITY_ACTION_TYPE_NONE , sa_out); +if (rc !=0 ) { +RTE_LOG (ERR ,USER1 ,"out bound create_sa failed, cfg\n" ); +return TEST_FAILED ; + } + + rc =create_sa (RTE_SECURITY_ACTION_TYPE_NONE , sa_in); +if (rc !=0 ) { +RTE_LOG (ERR ,USER1 ,"out bound create_sa failed, cfg\n" ); +return TEST_FAILED ; + } + +return TEST_SUCCESS ; +} + +static int +testsuite_setup (void ) +{ +struct rte_mbuf *mbuf[NUM_MBUF ]; +int ret, i; + + mbuf_pool =rte_pktmbuf_pool_create ("IPSEC_PERF_MBUFPOOL" , +NUM_MBUFS ,MBUF_CACHE_SIZE ,0 ,MBUF_SIZE , +rte_socket_id ()); +if (mbuf_pool ==NULL ) { +RTE_LOG (ERR ,USER1 ,"Can't create MBUFPOOL\n" ); +return TEST_FAILED ; + } + + cop_pool =rte_crypto_op_pool_create ( +"MBUF_CRYPTO_SYM_OP_POOL" , +RTE_CRYPTO_OP_TYPE_SYMMETRIC , +NUM_MBUFS ,MBUF_CACHE_SIZE , +DEFAULT_NUM_XFORMS * +sizeof (struct rte_crypto_sym_xform ) + +MAXIMUM_IV_LENGTH , +rte_socket_id ()); +if (cop_pool ==NULL ) { +RTE_LOG (ERR ,USER1 ,"Can't create CRYPTO_OP_POOL\n" ); +return TEST_FAILED ; + } + + ring_inb_prepare =rte_ring_create ("ring0" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_inb_prepare ==NULL ) +return TEST_FAILED ; + + ring_inb_process =rte_ring_create ("ring1" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_inb_process ==NULL ) +return TEST_FAILED ; + + ring_outb_prepare =rte_ring_create ("ring2" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_outb_prepare ==NULL ) +return TEST_FAILED ; + + ring_outb_process =rte_ring_create ("ring3" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_outb_process ==NULL ) +return TEST_FAILED ; + +for (i =0 ; i <NUM_MBUF ; i++) + mbuf[i] =generate_mbuf_data (mbuf_pool); + + ret =rte_ring_sp_enqueue_bulk (ring_inb_prepare, (void **)mbuf, +NUM_MBUF ,NULL ); +if (ret ==0 ) +return TEST_FAILED ; + +return TEST_SUCCESS ; +} + +static int +measure_performance (struct ipsec_sa *sa_out,struct ipsec_sa *sa_in) +{ +uint64_t time_diff =0 ; +uint64_t begin =0 ; +uint64_t hz =rte_get_timer_hz (); + + begin =rte_get_timer_cycles (); + +do { +if (create_traffic (sa_out, ring_inb_prepare, ring_inb_process, + ring_outb_prepare) <0 ) +return TEST_FAILED ; + +if (create_traffic (sa_in, ring_outb_prepare, ring_outb_process, + ring_inb_prepare) <0 ) +return TEST_FAILED ; + + time_diff =rte_get_timer_cycles () - begin; + + }while (time_diff < (hz *10 )); + +return TEST_SUCCESS ; +} + +static void +print_metrics (const struct ipsec_test_cfg *test_cfg, +struct ipsec_sa *sa_out,struct ipsec_sa *sa_in) +{ +printf ("\nMetrics of libipsec prepare/process api:\n" ); + +printf ("replay window size = %u\n" , test_cfg->replay_win_sz); +if (test_cfg->esn) +printf ("replay esn is enabled\n" ); +else +printf ("replay esn is disabled\n" ); +if (test_cfg->type ==RTE_CRYPTO_SYM_XFORM_AEAD ) +printf ("AEAD algo is AES_GCM\n" ); +else +printf ("CIPHER/AUTH algo is AES_CBC/SHA1\n" ); + + +printf ("avg cycles for a pkt prepare in outbound is = %.2Lf\n" , + (long double )sa_out->cnt.prepare_ticks_elapsed + / sa_out->cnt.nb_prepare_pkt); +printf ("avg cycles for a pkt process in outbound is = %.2Lf\n" , + (long double )sa_out->cnt.process_ticks_elapsed + / sa_out->cnt.nb_process_pkt); +printf ("avg cycles for a pkt prepare in inbound is = %.2Lf\n" , + (long double )sa_in->cnt.prepare_ticks_elapsed + / sa_in->cnt.nb_prepare_pkt); +printf ("avg cycles for a pkt process in inbound is = %.2Lf\n" , + (long double )sa_in->cnt.process_ticks_elapsed + / sa_in->cnt.nb_process_pkt); + +} + +static void +testsuite_teardown (void ) +{ +if (mbuf_pool !=NULL ) { +RTE_LOG (DEBUG ,USER1 ,"MBUFPOOL count %u\n" , +rte_mempool_avail_count (mbuf_pool)); +rte_mempool_free (mbuf_pool); + mbuf_pool =NULL ; + } + +if (cop_pool !=NULL ) { +RTE_LOG (DEBUG ,USER1 ,"CRYPTO_OP_POOL count %u\n" , +rte_mempool_avail_count (cop_pool)); +rte_mempool_free (cop_pool); + cop_pool =NULL ; + } + +rte_ring_free (ring_inb_prepare); +rte_ring_free (ring_inb_process); +rte_ring_free (ring_outb_prepare); +rte_ring_free (ring_outb_process); + + ring_inb_prepare =NULL ; + ring_inb_process =NULL ; + ring_outb_prepare =NULL ; + ring_outb_process =NULL ; +} + +static int +test_libipsec_perf (void ) +{ +struct ipsec_sa sa_out; +struct ipsec_sa sa_in; +int ret; + +if (testsuite_setup () <0 ) { +testsuite_teardown (); +return TEST_FAILED ; + } + +for (uint32_t i =0 ; i <RTE_DIM (test_cfg) ; i++) { + + ret =init_sa_session (&test_cfg[i], &sa_out, &sa_in); +if (ret !=0 ) { +testsuite_teardown (); +return TEST_FAILED ; + } + +if (measure_performance (&sa_out, &sa_in) <0 ) { +testsuite_teardown (); +return TEST_FAILED ; + } + +print_metrics (&test_cfg[i], &sa_out, &sa_in); + } + +testsuite_teardown (); + +return TEST_SUCCESS ; +} + +REGISTER_TEST_COMMAND (ipsec_perf_autotest, test_libipsec_perf);
Signature · 2 lines
Regards, Vladimir
Hi Savinay,
Quoted 40 lines
app/test/test_ipsec_perf.c new file +40
-0+static int +testsuite_setup (void ) +{ +struct rte_mbuf *mbuf[NUM_MBUF ]; +int ret, i; + + mbuf_pool =rte_pktmbuf_pool_create ("IPSEC_PERF_MBUFPOOL" , +NUM_MBUFS ,MBUF_CACHE_SIZE ,0 ,MBUF_SIZE , +rte_socket_id ()); +if (mbuf_pool ==NULL ) { +RTE_LOG (ERR ,USER1 ,"Can't create MBUFPOOL\n" ); +return TEST_FAILED ; + } + + cop_pool =rte_crypto_op_pool_create ( +"MBUF_CRYPTO_SYM_OP_POOL" , +RTE_CRYPTO_OP_TYPE_SYMMETRIC , +NUM_MBUFS ,MBUF_CACHE_SIZE , +DEFAULT_NUM_XFORMS * +sizeof (struct rte_crypto_sym_xform ) + +MAXIMUM_IV_LENGTH , +rte_socket_id ()); +if (cop_pool ==NULL ) { +RTE_LOG (ERR ,USER1 ,"Can't create CRYPTO_OP_POOL\n" ); +return TEST_FAILED ; + } + + ring_inb_prepare =rte_ring_create ("ring0" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_inb_prepare ==NULL ) +return TEST_FAILED ; + + ring_inb_process =rte_ring_create ("ring1" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_inb_process ==NULL ) +return TEST_FAILED ; + + ring_outb_prepare =rte_ring_create ("ring2" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_outb_prepare ==NULL )
app/test/test_ipsec_perf.c line 510
Quoted 10 lines
app/test/test_ipsec_perf.c new file +10
-0+return TEST_FAILED ; + + ring_outb_process =rte_ring_create ("ring3" ,RING_SIZE , +SOCKET_ID_ANY ,0 ); +if (ring_outb_process ==NULL ) +return TEST_FAILED ; + +for (i =0 ; i <NUM_MBUF ; i++) + mbuf[i] =generate_mbuf_data (mbuf_pool); +
I think enqueue below could be moved to the loop above, so don't need to allocate mbuf[NUM_MBUF] at stack.
struct rte_mbuf *mbuf;
...
for (i = 0; i < NUM_MBUF; i++) {
mbuf = generate_mbuf_data(mbuf_pool);
if (mbuf == NULL && rte_ring_sp_enqueue_bulk(ring_inb_prepare,
(void **)mbuf, 1) == 0) break; } If (i != NUM_BUF) return TEST_FAILED;
Quoted 8 lines
app/test/test_ipsec_perf.c new file +8
-0+ ret =rte_ring_sp_enqueue_bulk (ring_inb_prepare, (void **)mbuf, +NUM_MBUF ,NULL ); +if (ret ==0 ) +return TEST_FAILED ; + +return TEST_SUCCESS ; +} +
PATCH v3 1/1 app: test: measure libipsec performance
In reply to Savinay Dharmappatest app to measures the performance of libipsec api rte_ipsec_pkt_crypto and rte_ipsec_pkt_process.
--- MAINTAINERS | 2 + app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 618 +++++++++++++++++++++++++++++++++++++ 4 files changed, 623 insertions(+), 1 deletion(-) create mode 100644 app/test/test_ipsec_perf.c
MAINTAINERS +2 -0
@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/
app/test/Makefile +1 -1
@@ -231,7 +231,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BPF) += test_bpf.c SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsec
app/test/meson.build +2 -0
@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',
@@ -270,6 +271,7 @@ perf_test_names = [ 'rand_perf_autotest', 'hash_readwrite_perf_autotest', 'hash_readwrite_lf_perf_autotest', + 'ipsec_perf_autotest', ] driver_test_names = [
app/test/test_ipsec_perf.c new file +618 -0
618 lines not on this page · raw message
Signature · 1 line
2.17.1
Quoted 8 lines
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.Signed-off-by Savinay Dharmappa <[email protected]>---
I think the header need to be "app/test", or "test/ipsec". Apart from that:
Quoted 1 line
2.17.1
05/03/2020 12:45, Ananyev, Konstantin:
Quoted 1 line
I think the header need to be "app/test", or "test/ipsec".
It should be "test/ipsec" as it is an IPsec test command in the test application.
Quoted 2 lines
Apart from that:
I wonder why we have a different maintainer for each IPsec test. It looks a bit confusing.
IPsec - EXPERIMENTAL
M: Konstantin Ananyev <[email protected]>
T: git://dpdk.org/next/dpdk-next-crypto
F: lib/librte_ipsec/
M: Bernard Iremonger <[email protected]>
F: app/test/test_ipsec.c
F: doc/guides/prog_guide/ipsec_lib.rst
M: Savinay Dharmappa <[email protected]>
F: app/test/test_ipsec_perf.c
M: Vladimir Medvedkin <[email protected]>
F: app/test/test_ipsec_sad.c
F: app/test-sad/
In general, only one person takes the responsibility to be the main contact of any related question in the area.
So the file pattern could be:
F: app/test/test_ipsec*
Quoted 15 lines
05/03/2020 12:45, Ananyev, Konstantin:
Quoted 1 line
I think the header need to be "app/test", or "test/ipsec".
It should be "test/ipsec" as it is an IPsec test command in the test application.
Quoted 2 lines
Apart from that:
I wonder why we have a different maintainer for each IPsec test. It looks a bit confusing.
Quoted 12 lines
IPsec - EXPERIMENTAL
M: Konstantin Ananyev <[email protected]>
T:git://dpdk.org/next/dpdk-next-crypto
F:lib/librte_ipsec/
M: Bernard Iremonger <[email protected]>
F:app/test/test_ipsec.c
F:doc/guides/prog_guide/ipsec_lib.rst
M: Savinay Dharmappa <[email protected]>
F:app/test/test_ipsec_perf.c
M: Vladimir Medvedkin <[email protected]>
F:app/test/test_ipsec_sad.c
F: app/test-sad/
I suppose mainly by historical reasons -
each of these tests have different authors. Obviously each author claimed maintanership of his own code. Do you consider that as a problem?
Quoted 7 lines
In general, only one person takes the responsibility to be the main contact of any related question in the area.
So the file pattern could be:
F: app/test/test_ipsec*
05/03/2020 23:51, Ananyev, Konstantin:
Quoted 26 lines
Quoted 13 lines
05/03/2020 12:45, Ananyev, Konstantin:
Quoted 1 line
I think the header need to be "app/test", or "test/ipsec".
It should be "test/ipsec" as it is an IPsec test command in the test application.
Quoted 2 lines
Apart from that:
I wonder why we have a different maintainer for each IPsec test. It looks a bit confusing.
Quoted 12 lines
IPsec - EXPERIMENTAL
M: Konstantin Ananyev <[email protected]>
T:git://dpdk.org/next/dpdk-next-crypto
F:lib/librte_ipsec/
M: Bernard Iremonger <[email protected]>
F:app/test/test_ipsec.c
F:doc/guides/prog_guide/ipsec_lib.rst
M: Savinay Dharmappa <[email protected]>
F:app/test/test_ipsec_perf.c
M: Vladimir Medvedkin <[email protected]>
F:app/test/test_ipsec_sad.c
F: app/test-sad/
Quoted 4 lines
I suppose mainly by historical reasons -
each of these tests have different authors. Obviously each author claimed maintanership of his own code. Do you consider that as a problem?
Yes I consider a maintainer should know a whole area
and be the point of contact for this area.
For instance, when there is a bug with a test,
we don't know whether the problem is in the test or in the library.
Note that the original author can be found with the git history.
Quoted 5 lines
Quoted 5 lines
In general, only one person takes the responsibility to be the main contact of any related question in the area.
So the file pattern could be:
F: app/test/test_ipsec*
Quoted 33 lines
05/03/2020 23:51, Ananyev, Konstantin:
Quoted 26 lines
Quoted 13 lines
05/03/2020 12:45, Ananyev, Konstantin:
Quoted 1 line
I think the header need to be "app/test", or "test/ipsec".
It should be "test/ipsec" as it is an IPsec test command in the test application.
Quoted 2 lines
Apart from that:
I wonder why we have a different maintainer for each IPsec test. It looks a bit confusing.
Quoted 12 lines
IPsec - EXPERIMENTAL
M: Konstantin Ananyev <[email protected]>
T:git://dpdk.org/next/dpdk-next-crypto
F:lib/librte_ipsec/
M: Bernard Iremonger <[email protected]>
F:app/test/test_ipsec.c
F:doc/guides/prog_guide/ipsec_lib.rst
M: Savinay Dharmappa <[email protected]>
F:app/test/test_ipsec_perf.c
M: Vladimir Medvedkin <[email protected]>
F:app/test/test_ipsec_sad.c
F: app/test-sad/Quoted 4 lines
I suppose mainly by historical reasons -
each of these tests have different authors. Obviously each author claimed maintanership of his own code. Do you consider that as a problem?
Quoted 4 lines
Yes I consider a maintainer should know a whole area
and be the point of contact for this area.
For instance, when there is a bug with a test,
we don't know whether the problem is in the test or in the library.
But same problem could appear with any test/sample app.
Let say you have a problem with testpmd.
Is that testpmd itself, rte_ethdev or particular PMD? Obviously some preliminary investigation is needed before finger pointing. I don't see why having multiple persons as maintainers for the same area is a problem. In fact, it probably gives more chances for faster turnover (as long as they do their maintainer job properly).
Quoted 10 lines
Note that the original author can be found with the git history.
Quoted 5 lines
Quoted 5 lines
In general, only one person takes the responsibility to be the main contact of any related question in the area.
So the file pattern could be:
F: app/test/test_ipsec*
10/03/2020 14:24, Ananyev, Konstantin:
Quoted 38 lines
Quoted 33 lines
05/03/2020 23:51, Ananyev, Konstantin:
Quoted 26 lines
Quoted 13 lines
05/03/2020 12:45, Ananyev, Konstantin:
Quoted 1 line
I think the header need to be "app/test", or "test/ipsec".
It should be "test/ipsec" as it is an IPsec test command in the test application.
Quoted 2 lines
Apart from that:
I wonder why we have a different maintainer for each IPsec test. It looks a bit confusing.
Quoted 12 lines
IPsec - EXPERIMENTAL
M: Konstantin Ananyev <[email protected]>
T:git://dpdk.org/next/dpdk-next-crypto
F:lib/librte_ipsec/
M: Bernard Iremonger <[email protected]>
F:app/test/test_ipsec.c
F:doc/guides/prog_guide/ipsec_lib.rst
M: Savinay Dharmappa <[email protected]>
F:app/test/test_ipsec_perf.c
M: Vladimir Medvedkin <[email protected]>
F:app/test/test_ipsec_sad.c
F: app/test-sad/Quoted 4 lines
I suppose mainly by historical reasons -
each of these tests have different authors. Obviously each author claimed maintanership of his own code. Do you consider that as a problem?Quoted 4 lines
Yes I consider a maintainer should know a whole area
and be the point of contact for this area.
For instance, when there is a bug with a test,
we don't know whether the problem is in the test or in the library.
Quoted 7 lines
But same problem could appear with any test/sample app.
Let say you have a problem with testpmd.
Is that testpmd itself,rte_ethdevor particular PMD? Obviously some preliminary investigation is needed before finger pointing. I don't see why having multiple persons as maintainers for the same area is a problem. In fact, it probably gives more chances for faster turnover (as long as they do their maintainer job properly).
Yes, having several maintainers is good.
But having one different maintainer per test file is quite ridiculous. In general the maintainers of a lib are also maintaining the related app files.
Quoted 11 lines
Quoted 9 lines
Note that the original author can be found with the git history.
Quoted 5 lines
Quoted 5 lines
In general, only one person takes the responsibility to be the main contact of any related question in the area.
So the file pattern could be:
F: app/test/test_ipsec*
PATCH v4 1/1 test/ipsec: measure libipsec performance
In reply to Savinay Dharmappatest app to measures the performance of libipsec api rte_ipsec_pkt_crypto and rte_ipsec_pkt_process.
--- MAINTAINERS | 2 + app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 619 +++++++++++++++++++++++++++++++++++++ 4 files changed, 624 insertions(+), 1 deletion(-) create mode 100644 app/test/test_ipsec_perf.c
MAINTAINERS +2 -0
@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/
app/test/Makefile +1 -1
@@ -231,7 +231,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_BPF) += test_bpf.c SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsec
app/test/meson.build +2 -0
@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',
@@ -270,6 +271,7 @@ perf_test_names = [ 'rand_perf_autotest', 'hash_readwrite_perf_autotest', 'hash_readwrite_lf_perf_autotest', + 'ipsec_perf_autotest', ] driver_test_names = [
app/test/test_ipsec_perf.c new file +619 -0
619 lines not on this page · raw message
Signature · 1 line
2.17.1
Akhil Goyal
AppliedRe: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
Quoted 6 lines
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.Signed-off-by Savinay Dharmappa <[email protected]>---
Applied to dpdk-next-crypto
Thanks.
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
In reply to Savinay Dharmappa06/03/2020 08:08, Savinay Dharmappa:
Quoted 2 lines
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.
Please make sentences in commit logs.
Quoted 6 lines
Signed-off-by Savinay Dharmappa <[email protected]>---
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 8 lines
MAINTAINERS +2
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/
Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
21/04/2020 04:29, Thomas Monjalon:
Quoted 16 lines
06/03/2020 08:08, Savinay Dharmappa:
Quoted 2 lines
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.Please make sentences in commit logs.
Quoted 10 lines
Signed-off-by Savinay Dharmappa <[email protected]>---
MAINTAINERS +1
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]>
Quoted 8 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 4 lines
MAINTAINERS +1
-0+F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.
In addition, reading the first lines of the file, I see two mistakes:
app/test/test_ipsec_perf.c line 2
Quoted 5 lines
app/test/test_ipsec_perf.c new file +2
-0@@ -0,0 +1,619 @@ +/* SPDX-License-Identifier: BSD-3-Clause +* Copyright(c) 2010-2014 Intel Corporation
The Copyright year should be 2020 I guess.
app/test/test_ipsec_perf.c line 10
Quoted 8 lines
app/test/test_ipsec_perf.c new file +8
-0+*/ + +#include <stdio.h> +#include <rte_ip.h> +#include <rte_malloc.h> +#include <rte_ring.h> +#include <rte_mbuf.h> +#include <rte_malloc.h>
rte_malloc.h is already included 3 lines above.
I have a bad feeling about this patch,
I think it should be dropped from 20.05-rc1.
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
Quoted 51 lines
21/04/2020 04:29, Thomas Monjalon:
Quoted 16 lines
06/03/2020 08:08, Savinay Dharmappa:
Quoted 2 lines
test app to measures the performance of libipsec api
rte_ipsec_pkt_cryptoandrte_ipsec_pkt_process.Please make sentences in commit logs.
Quoted 10 lines
Signed-off-by Savinay Dharmappa <[email protected]>---
MAINTAINERS +1
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]>Quoted 8 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 4 lines
MAINTAINERS +1
-0+F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.In addition, reading the first lines of the file, I see two mistakes:
app/test/test_ipsec_perf.c line 2
Quoted 5 lines
app/test/test_ipsec_perf.c new file +2
-0@@ -0,0 +1,619 @@ +/* SPDX-License-Identifier: BSD-3-Clause +* Copyright(c) 2010-2014 Intel Corporation The Copyright year should be 2020 I guess.
app/test/test_ipsec_perf.c line 10
Quoted 8 lines
app/test/test_ipsec_perf.c new file +8
-0+*/ + +#include <stdio.h> +#include <rte_ip.h> +#include <rte_malloc.h> +#include <rte_ring.h> +#include <rte_mbuf.h> +#include <rte_malloc.h>
rte_malloc.his already included 3 lines above.
Quoted 2 lines
I have a bad feeling about this patch,
I think it should be dropped from 20.05-rc1.
If you feel patch is not good enough - sure feel free to drop it, we'll try to fix the issues and resubmit it for rc2. About having separate MAINTAINER for the test - honestly I don't understand why it is a problem for you. Obviously we would like to spread the load - what's wrong with it? Konstantin
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
21/04/2020 12:21, Ananyev, Konstantin:
Quoted 16 lines
Quoted 16 lines
21/04/2020 04:29, Thomas Monjalon:
Quoted 15 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 11 lines
MAINTAINERS +2
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.
[...]
Quoted 3 lines
About having separate MAINTAINER for the test - honestly I don't understand why it is a problem for you. Obviously we would like to spread the load - what's wrong with it?
This is a problem of ownership.
Maintaining a library means you take care of every aspect, including tests. That's why I would like to see you as a global maintainer of IPsec.
It doesn't prevent you to delegate workload, of course.
But at the end it is more convenient to know there is a limited number of persons responsible for the global quality of a component, a person which is accountable and answering questions on the topic, no matter which exact file we are talking about.
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
Quoted 27 lines
21/04/2020 12:21, Ananyev, Konstantin:
Quoted 16 lines
Quoted 16 lines
21/04/2020 04:29, Thomas Monjalon:
Quoted 15 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 11 lines
MAINTAINERS +2
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.[...]
Quoted 3 lines
About having separate MAINTAINER for the test - honestly I don't understand why it is a problem for you. Obviously we would like to spread the load - what's wrong with it?
This is a problem of ownership.
Maintaining a library means you take care of every aspect, including tests. That's why I would like to see you as a global maintainer of IPsec.
Quoted 5 lines
It doesn't prevent you to delegate workload, of course.
But at the end it is more convenient to know there is a limited number of persons responsible for the global quality of a component, a person which is accountable and answering questions on the topic, no matter which exact file we are talking about.
Just talked with Bernard, he kindly agreed to be a maintainer for all ipsec UT:
app/test/test_ipsec*
Hope that will fulfil your concern?
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
21/04/2020 13:07, Ananyev, Konstantin:
Quoted 32 lines
Quoted 26 lines
21/04/2020 12:21, Ananyev, Konstantin:
Quoted 16 lines
Quoted 16 lines
21/04/2020 04:29, Thomas Monjalon:
Quoted 15 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 11 lines
MAINTAINERS +2
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.[...]
Quoted 3 lines
About having separate MAINTAINER for the test - honestly I don't understand why it is a problem for you. Obviously we would like to spread the load - what's wrong with it?
This is a problem of ownership.
Maintaining a library means you take care of every aspect, including tests. That's why I would like to see you as a global maintainer of IPsec.Quoted 5 lines
It doesn't prevent you to delegate workload, of course.
But at the end it is more convenient to know there is a limited number of persons responsible for the global quality of a component, a person which is accountable and answering questions on the topic, no matter which exact file we are talking about.
Quoted 3 lines
Just talked with Bernard, he kindly agreed to be a maintainer for all ipsec UT:
app/test/test_ipsec*
Hope that will fulfil your concern?
My concern was to have the library maintainer maintaining also the related tests. I don't understand why you don't want to take this responsibility, but I cannot force you.
Having only one maintainer for IPsec tests is better than the current situation.
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
Quoted 33 lines
21/04/2020 13:07, Ananyev, Konstantin:
Quoted 32 lines
Quoted 31 lines
21/04/2020 12:21, Ananyev, Konstantin:
Quoted 16 lines
Quoted 16 lines
21/04/2020 04:29, Thomas Monjalon:
Quoted 15 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 11 lines
MAINTAINERS +2
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.[...]
Quoted 3 lines
About having separate MAINTAINER for the test - honestly I don't understand why it is a problem for you. Obviously we would like to spread the load - what's wrong with it?
This is a problem of ownership.
Maintaining a library means you take care of every aspect, including tests. That's why I would like to see you as a global maintainer of IPsec.It doesn't prevent you to delegate workload, of course.
But at the end it is more convenient to know there is a limited number of persons responsible for the global quality of a component, a person which is accountable and answering questions on the topic, no matter which exact file we are talking about.
Quoted 8 lines
Quoted 3 lines
Just talked with Bernard, he kindly agreed to be a maintainer for all ipsec UT:
app/test/test_ipsec*
Hope that will fulfil your concern?My concern was to have the library maintainer maintaining also the related tests. I don't understand why you don't want to take this responsibility, but I cannot force you.
Having only one maintainer for IPsec tests is better than the current situation.
Ok, are you going to drop v4, so we can submit v5 with the fixes? Or should we submit a patch with fixes on top of v4? Konstantin
Re: [dpdk-dev] [PATCH v4] test/ipsec: measure libipsec performance
21/04/2020 14:04, Ananyev, Konstantin:
Quoted 37 lines
Quoted 37 lines
21/04/2020 13:07, Ananyev, Konstantin:
Quoted 35 lines
Quoted 26 lines
21/04/2020 12:21, Ananyev, Konstantin:
Quoted 16 lines
Quoted 16 lines
21/04/2020 04:29, Thomas Monjalon:
Quoted 15 lines
MAINTAINERS F: lib/librte_ipsec/ · line 1266
Quoted 11 lines
MAINTAINERS +2
-0@@ -1259,6 +1259,8 @@ F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> F: app/test/test_ipsec.c F: doc/guides/prog_guide/ipsec_lib.rst +M: Savinay Dharmappa <[email protected]> +F: app/test/test_ipsec_perf.c M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c F: app/test-sad/Repeating what I said on v3:
Having one different maintainer per test file is quite ridiculous. The maintainers of a lib are expected to maintain the related tests.[...]
Quoted 3 lines
About having separate MAINTAINER for the test - honestly I don't understand why it is a problem for you. Obviously we would like to spread the load - what's wrong with it?
This is a problem of ownership.
Maintaining a library means you take care of every aspect, including tests. That's why I would like to see you as a global maintainer of IPsec.Quoted 5 lines
It doesn't prevent you to delegate workload, of course.
But at the end it is more convenient to know there is a limited number of persons responsible for the global quality of a component, a person which is accountable and answering questions on the topic, no matter which exact file we are talking about.Just talked with Bernard, he kindly agreed to be a maintainer for all ipsec UT:
app/test/test_ipsec*
Hope that will fulfil your concern?
Quoted 7 lines
Quoted 4 lines
My concern was to have the library maintainer maintaining also the related tests. I don't understand why you don't want to take this responsibility, but I cannot force you.
Having only one maintainer for IPsec tests is better than the current situation.Ok, are you going to drop v4, so we can submit v5 with the fixes? Or should we submit a patch with fixes on top of v4?
v4 is dropped, you can send a v5.
PATCH v5 1/1 test/ipsec: measure libipsec performance
In reply to Savinay DharmappaAdd new test-case to measure performance of
ipsec data-path functions.
--- MAINTAINERS | 2 +- app/test/Makefile | 2 +- app/test/meson.build | 2 + app/test/test_ipsec_perf.c | 614 +++++++++++++++++++++++++++++++++++++ 4 files changed, 618 insertions(+), 2 deletions(-) create mode 100644 app/test/test_ipsec_perf.c
MAINTAINERS +1 -1
@@ -1251,7 +1251,7 @@ M: Konstantin Ananyev <[email protected]> T: git://dpdk.org/next/dpdk-next-crypto F: lib/librte_ipsec/ M: Bernard Iremonger <[email protected]> -F: app/test/test_ipsec.c +F: app/test/test_ipsec* F: doc/guides/prog_guide/ipsec_lib.rst M: Vladimir Medvedkin <[email protected]> F: app/test/test_ipsec_sad.c
app/test/Makefile +1 -1
@@ -240,7 +240,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_RCU) += test_rcu_qsbr.c test_rcu_qsbr_perf.c SRCS-$(CONFIG_RTE_LIBRTE_SECURITY) += test_security.c -SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c +SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec.c test_ipsec_perf.c SRCS-$(CONFIG_RTE_LIBRTE_IPSEC) += test_ipsec_sad.c ifeq ($(CONFIG_RTE_LIBRTE_IPSEC),y) LDLIBS += -lrte_ipsec
app/test/meson.build +2 -0
@@ -60,6 +60,7 @@ test_sources = files('commands.c', 'test_interrupts.c', 'test_ipsec.c', 'test_ipsec_sad.c', + 'test_ipsec_perf.c', 'test_kni.c', 'test_kvargs.c', 'test_link_bonding.c',
@@ -285,6 +286,7 @@ perf_test_names = [ 'hash_readwrite_perf_autotest', 'hash_readwrite_lf_perf_autotest', 'trace_perf_autotest', + 'ipsec_perf_autotest', ] driver_test_names = [
app/test/test_ipsec_perf.c new file +614 -0
614 lines not on this page · raw message
Signature · 1 line
2.17.1
Re: [dpdk-dev] [PATCH v5] test/ipsec: measure libipsec performance
Quoted 8 lines
-----Original Message-----
From: Dharmappa, Savinay <[email protected]>
Sent: Thursday, April 23, 2020 4:25 PM
To: [email protected]; Ananyev, Konstantin <[email protected]>; [email protected]
Subject: [PATCH v5] test/ipsec: measure libipsec performance
Quoted 5 lines
Add new test-case to measure performance of
ipsec data-path functions.Signed-off-by Savinay Dharmappa <[email protected]>---
Quoted 2 lines
--
2.17.1
Akhil Goyal
AppliedRe: [dpdk-dev] [PATCH v5] test/ipsec: measure libipsec performance
Quoted 11 lines
Quoted 7 lines
Add new test-case to measure performance of
ipsec data-path functions.Signed-off-by Savinay Dharmappa <[email protected]>---
Tested-by Konstantin Ananyev <[email protected]>
Applied to dpdk-next-crypto