ivampiresp/Spdk
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
a497ee9b56
Spdk/test/unit/lib/nvme/nvme_tcp.c/nvme_tcp_ut.c

824 lines
28 KiB
C
Raw Normal View History

nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
/*-
* BSD LICENSE
*
* Copyright (c) Intel Corporation.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "spdk/stdinc.h"
#include "spdk_cunit.h"
#include "common/lib/test_sock.c"
#include "nvme/nvme_tcp.c"
test/nvme: add common_stubs.h file. A lot of the stubs between these transports will end up being common. Change-Id: Ib9c8ff947b95f34633eb13953405d3153a7f4ac7 Signed-off-by: Seth Howell <seth.howell@intel.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/479602 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2020-01-07 21:18:38 +00:00
#include "common/lib/nvme/common_stubs.h"
nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
log: simplify SPDK_LOG_REGISTER_COMPONENT This patch removes the string from register component. Removed are all instances in libs or hardcoded in apps. Starting with this patch literal passed to register, serves as name for the flag. All instances of SPDK_LOG_* were replaced with just * in lowercase. No actual name change for flags occur in this patch. Affected are SPDK_LOG_REGISTER_COMPONENT() and SPDK_*LOG() macros. Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com> Change-Id: I002b232fde57ecf9c6777726b181fc0341f1bb17 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/4495 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Mellanox Build Bot Reviewed-by: Anil Veerabhadrappa <anil.veerabhadrappa@broadcom.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Community-CI: Broadcom CI
2020-09-04 11:27:29 +00:00
SPDK_LOG_REGISTER_COMPONENT(nvme)
nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
DEFINE_STUB(nvme_qpair_submit_request,
int, (struct spdk_nvme_qpair *qpair, struct nvme_request *req), 0);
nvme: Add the priority field in struct spdk_nvme_transport_id Purpose: To set the priority of the NVMe-oF connection especially for TCP connection. For example, the previous example can be: trtype:TCP adrfam:IPv4 traddr:10.67.110.181 trsvcid:4420 With the change, it could be: trtype:TCP adrfam:IPv4 traddr:10.67.110.181 trsvcid:4420 priority:2 The priority is optional. We try to change spdk_nvme_transport_id but not in spdk_nvme_ctrlr_opts since the opts in spdk_nvme_ctrlr_opts will reflect in every nvme ctrlr, this is short of flexibility. Signed-off-by: Ziye Yang <ziye.yang@intel.com> Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com> Change-Id: I1ba364c714a95f2dbeab2b3fcc832b0222b48a15 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1875 Community-CI: Mellanox Build Bot Community-CI: Broadcom CI Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
2020-02-19 11:18:51 +00:00
DEFINE_STUB(spdk_sock_set_priority,
int, (struct spdk_sock *sock, int priority), 0);
nvme: implement epoll in the tcp transport. Change-Id: I6672361baca4969f23259c19b73ed9dbe2f436bd Signed-off-by: Seth Howell <seth.howell@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/885 Community-CI: Mellanox Build Bot Community-CI: Broadcom CI Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2020-02-13 21:52:59 +00:00
DEFINE_STUB(spdk_nvme_poll_group_remove, int, (struct spdk_nvme_poll_group *group,
struct spdk_nvme_qpair *qpair), 0);
nvme/tcp: Add the implementation to get the optimal polling group Add the real support in nvme tcp transport. Change-Id: I2aa9b0284d6fe009925e67f602a055e787f77987 Signed-off-by: Ziye Yang <ziye.yang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5734 Community-CI: Broadcom CI Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2020-07-13 15:48:29 +00:00
DEFINE_STUB(spdk_sock_get_optimal_sock_group,
int,
(struct spdk_sock *sock, struct spdk_sock_group **group),
0);
DEFINE_STUB(spdk_sock_group_get_ctx,
void *,
(struct spdk_sock_group *group),
NULL);
nvme: implement epoll in the tcp transport. Change-Id: I6672361baca4969f23259c19b73ed9dbe2f436bd Signed-off-by: Seth Howell <seth.howell@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/885 Community-CI: Mellanox Build Bot Community-CI: Broadcom CI Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2020-02-13 21:52:59 +00:00
nvme/tcp: Process poll_group when waiting for icresp A preparation step for enabling zero copy in NVMEoF TCP initiator. Since nvme_tcp_qpair_process_completions doesn't process poll group, we can't get asycn notification from kernel. 1. Add a qpair to poll group before we send icreq in order to be able to process buffer reclaim notification. 2. Check if qpair is connected to a poll group and call nvme_tcp_poll_group_process_completions instead of nvme_tcp_qpair_process_completions when waiting for icresp 3. Add processing of poll group to nvme_wait_for_completion_timeout and nvme_wait_for_completion_robust_lock since they are used to process FABRIC_CONNECT command Change-Id: I38d2d9496bca8d0cd72e44883df2df802e31a87d Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/4208 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2020-09-01 08:37:36 +00:00
DEFINE_STUB(spdk_nvme_poll_group_process_completions, int64_t, (struct spdk_nvme_poll_group *group,
uint32_t completions_per_qpair, spdk_nvme_disconnected_qpair_cb disconnected_qpair_cb), 0);
DEFINE_STUB(nvme_poll_group_connect_qpair, int, (struct spdk_nvme_qpair *qpair), 0);
nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
static void
test_nvme_tcp_pdu_set_data_buf(void)
{
struct nvme_tcp_pdu pdu = {};
struct iovec iov[NVME_TCP_MAX_SGL_DESCRIPTORS] = {};
uint32_t data_len;
uint64_t i;
/* 1st case: input is a single SGL entry. */
iov[0].iov_base = (void *)0xDEADBEEF;
iov[0].iov_len = 4096;
nvme_tcp_pdu_set_data_buf(&pdu, iov, 1, 1024, 512);
CU_ASSERT(pdu.data_iovcnt == 1);
CU_ASSERT((uint64_t)pdu.data_iov[0].iov_base == 0xDEADBEEF + 1024);
CU_ASSERT(pdu.data_iov[0].iov_len == 512);
/* 2nd case: simulate split on multiple SGL entries. */
iov[0].iov_base = (void *)0xDEADBEEF;
iov[0].iov_len = 4096;
iov[1].iov_base = (void *)0xFEEDBEEF;
iov[1].iov_len = 512 * 7;
iov[2].iov_base = (void *)0xF00DF00D;
iov[2].iov_len = 4096 * 2;
nvme_tcp_pdu_set_data_buf(&pdu, iov, 3, 0, 2048);
CU_ASSERT(pdu.data_iovcnt == 1);
CU_ASSERT((uint64_t)pdu.data_iov[0].iov_base == 0xDEADBEEF);
CU_ASSERT(pdu.data_iov[0].iov_len == 2048);
nvme_tcp_pdu_set_data_buf(&pdu, iov, 3, 2048, 2048 + 512 * 3);
CU_ASSERT(pdu.data_iovcnt == 2);
CU_ASSERT((uint64_t)pdu.data_iov[0].iov_base == 0xDEADBEEF + 2048);
CU_ASSERT(pdu.data_iov[0].iov_len == 2048);
CU_ASSERT((uint64_t)pdu.data_iov[1].iov_base == 0xFEEDBEEF);
CU_ASSERT(pdu.data_iov[1].iov_len == 512 * 3);
nvme_tcp_pdu_set_data_buf(&pdu, iov, 3, 4096 + 512 * 3, 512 * 4 + 4096 * 2);
CU_ASSERT(pdu.data_iovcnt == 2);
CU_ASSERT((uint64_t)pdu.data_iov[0].iov_base == 0xFEEDBEEF + 512 * 3);
CU_ASSERT(pdu.data_iov[0].iov_len == 512 * 4);
CU_ASSERT((uint64_t)pdu.data_iov[1].iov_base == 0xF00DF00D);
CU_ASSERT(pdu.data_iov[1].iov_len == 4096 * 2);
/* 3rd case: Number of input SGL entries is equal to the number of PDU SGL
* entries.
*/
data_len = 0;
for (i = 0; i < NVME_TCP_MAX_SGL_DESCRIPTORS; i++) {
iov[i].iov_base = (void *)(0xDEADBEEF + i);
iov[i].iov_len = 512 * (i + 1);
data_len += 512 * (i + 1);
}
nvme_tcp_pdu_set_data_buf(&pdu, iov, NVME_TCP_MAX_SGL_DESCRIPTORS, 0, data_len);
CU_ASSERT(pdu.data_iovcnt == NVME_TCP_MAX_SGL_DESCRIPTORS);
for (i = 0; i < NVME_TCP_MAX_SGL_DESCRIPTORS; i++) {
CU_ASSERT((uint64_t)pdu.data_iov[i].iov_base == 0xDEADBEEF + i);
CU_ASSERT(pdu.data_iov[i].iov_len == 512 * (i + 1));
}
}
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
static void
test_nvme_tcp_build_iovs(void)
{
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
const uintptr_t pdu_iov_len = 4096;
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
struct nvme_tcp_pdu pdu = {};
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
struct iovec iovs[5] = {};
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
uint32_t mapped_length = 0;
int rc;
Revert "nvme/tcp: Change hdr in nvme_tcp_pdu to pointer" This reverts commit ea5ad0b28630bf735b3efbb126c0ac00b0403647. This code is moving from the nvmf target to the posix sock layer in this series. Change-Id: I333bdf325848e726ab82a9e6916e1bbdcd34009c Signed-off-by: Ben Walker <benjamin.walker@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/446 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2019-09-27 21:19:57 +00:00
pdu.hdr.common.pdu_type = SPDK_NVME_TCP_PDU_TYPE_CAPSULE_CMD;
pdu.hdr.common.hlen = sizeof(struct spdk_nvme_tcp_cmd);
pdu.hdr.common.plen = pdu.hdr.common.hlen + SPDK_NVME_TCP_DIGEST_LEN + pdu_iov_len * 2 +
SPDK_NVME_TCP_DIGEST_LEN;
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
pdu.data_len = pdu_iov_len * 2;
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
pdu.padding_len = 0;
pdu.data_iov[0].iov_base = (void *)0xDEADBEEF;
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
pdu.data_iov[0].iov_len = pdu_iov_len;
pdu.data_iov[1].iov_base = (void *)(0xDEADBEEF + pdu_iov_len);
pdu.data_iov[1].iov_len = pdu_iov_len;
pdu.data_iovcnt = 2;
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
rc = nvme_tcp_build_iovs(iovs, 5, &pdu, true, true, &mapped_length);
CU_ASSERT(rc == 4);
Revert "nvme/tcp: Change hdr in nvme_tcp_pdu to pointer" This reverts commit ea5ad0b28630bf735b3efbb126c0ac00b0403647. This code is moving from the nvmf target to the posix sock layer in this series. Change-Id: I333bdf325848e726ab82a9e6916e1bbdcd34009c Signed-off-by: Ben Walker <benjamin.walker@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/446 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2019-09-27 21:19:57 +00:00
CU_ASSERT(iovs[0].iov_base == (void *)&pdu.hdr.raw);
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
CU_ASSERT(iovs[0].iov_len == sizeof(struct spdk_nvme_tcp_cmd) + SPDK_NVME_TCP_DIGEST_LEN);
CU_ASSERT(iovs[1].iov_base == (void *)0xDEADBEEF);
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
CU_ASSERT(iovs[1].iov_len == pdu_iov_len);
CU_ASSERT(iovs[2].iov_base == (void *)(0xDEADBEEF + pdu_iov_len));
CU_ASSERT(iovs[2].iov_len == pdu_iov_len);
CU_ASSERT(iovs[3].iov_base == (void *)pdu.data_digest);
CU_ASSERT(iovs[3].iov_len == SPDK_NVME_TCP_DIGEST_LEN);
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
CU_ASSERT(mapped_length == sizeof(struct spdk_nvme_tcp_cmd) + SPDK_NVME_TCP_DIGEST_LEN +
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
pdu_iov_len * 2 + SPDK_NVME_TCP_DIGEST_LEN);
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
/* Add a new data_iov entry, update pdu iov count and data length */
pdu.data_iov[2].iov_base = (void *)(0xBAADF00D);
pdu.data_iov[2].iov_len = 123;
pdu.data_iovcnt = 3;
pdu.data_len += 123;
Revert "nvme/tcp: Change hdr in nvme_tcp_pdu to pointer" This reverts commit ea5ad0b28630bf735b3efbb126c0ac00b0403647. This code is moving from the nvmf target to the posix sock layer in this series. Change-Id: I333bdf325848e726ab82a9e6916e1bbdcd34009c Signed-off-by: Ben Walker <benjamin.walker@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/446 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2019-09-27 21:19:57 +00:00
pdu.hdr.common.plen += 123;
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
rc = nvme_tcp_build_iovs(iovs, 5, &pdu, true, true, &mapped_length);
CU_ASSERT(rc == 5);
Revert "nvme/tcp: Change hdr in nvme_tcp_pdu to pointer" This reverts commit ea5ad0b28630bf735b3efbb126c0ac00b0403647. This code is moving from the nvmf target to the posix sock layer in this series. Change-Id: I333bdf325848e726ab82a9e6916e1bbdcd34009c Signed-off-by: Ben Walker <benjamin.walker@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/446 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2019-09-27 21:19:57 +00:00
CU_ASSERT(iovs[0].iov_base == (void *)&pdu.hdr.raw);
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
CU_ASSERT(iovs[0].iov_len == sizeof(struct spdk_nvme_tcp_cmd) + SPDK_NVME_TCP_DIGEST_LEN);
CU_ASSERT(iovs[1].iov_base == (void *)0xDEADBEEF);
tcp: Cleanup nvme_tcp_pdu structure With the recent changes which added usage of writev_async to both TCP target and initiator, nvme_tcp_pdu::writev_offset becomes useless since it is not updated in data path. This field is only used in UT. Remove this field from nvme_tcp_pdu structure, now nvme_tcp_build_iovs builds iov which fully describes the PDU. Update UT accordingly. Field padding_valid_bytes is not used at all, delete it too Change-Id: I2d6040ae64d6847cb455f59f65ec5677de8e5192 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/483374 Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
2020-01-30 12:23:08 +00:00
CU_ASSERT(iovs[1].iov_len == pdu_iov_len);
CU_ASSERT(iovs[2].iov_base == (void *)(0xDEADBEEF + pdu_iov_len));
CU_ASSERT(iovs[2].iov_len == pdu_iov_len);
CU_ASSERT(iovs[3].iov_base == (void *)(0xBAADF00D));
CU_ASSERT(iovs[3].iov_len == 123);
CU_ASSERT(iovs[4].iov_base == (void *)pdu.data_digest);
CU_ASSERT(iovs[4].iov_len == SPDK_NVME_TCP_DIGEST_LEN);
CU_ASSERT(mapped_length == sizeof(struct spdk_nvme_tcp_cmd) + SPDK_NVME_TCP_DIGEST_LEN +
pdu_iov_len * 2 + SPDK_NVME_TCP_DIGEST_LEN + 123);
nvme/tcp: Add UT to nvme_tcp_build_iovs There was no UT to process SGL for PDU, and add UT in this patch. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: Icd07683ba94b584cd3c6e5a88fee78d512d5832d Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458541 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Ziye Yang <ziye.yang@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com>
2019-06-19 04:20:26 +00:00
}
unit/nvme: Add unit test for the tcp sgl It seems that autotest doesn't cover the function nvme_tcp_build_sgl_request(), So add the unit test. Change-Id: I2e243910cb38ffc49bdd6048b365dc7fa60892c2 Signed-off-by: JinYu <jin.yu@intel.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456728 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-06-04 10:47:09 +00:00
struct nvme_tcp_ut_bdev_io {
struct iovec iovs[NVME_TCP_MAX_SGL_DESCRIPTORS];
int iovpos;
};
/* essentially a simplification of bdev_nvme_next_sge and bdev_nvme_reset_sgl */
static void
nvme_tcp_ut_reset_sgl(void *cb_arg, uint32_t offset)
{
struct nvme_tcp_ut_bdev_io *bio = cb_arg;
struct iovec *iov;
for (bio->iovpos = 0; bio->iovpos < NVME_TCP_MAX_SGL_DESCRIPTORS; bio->iovpos++) {
iov = &bio->iovs[bio->iovpos];
/* Offset must be aligned with the start of any SGL entry */
if (offset == 0) {
break;
}
SPDK_CU_ASSERT_FATAL(offset >= iov->iov_len);
offset -= iov->iov_len;
}
SPDK_CU_ASSERT_FATAL(offset == 0);
SPDK_CU_ASSERT_FATAL(bio->iovpos < NVME_TCP_MAX_SGL_DESCRIPTORS);
}
static int
nvme_tcp_ut_next_sge(void *cb_arg, void **address, uint32_t *length)
{
struct nvme_tcp_ut_bdev_io *bio = cb_arg;
struct iovec *iov;
SPDK_CU_ASSERT_FATAL(bio->iovpos < NVME_TCP_MAX_SGL_DESCRIPTORS);
iov = &bio->iovs[bio->iovpos];
*address = iov->iov_base;
*length = iov->iov_len;
bio->iovpos++;
return 0;
}
static void
test_nvme_tcp_build_sgl_request(void)
{
struct nvme_tcp_qpair tqpair;
lib/nvme_tcp: get the max_sges from the nvme ctrlr. Add the error print if there is still remaining_size in order to provide more meaningful debug info. Signed-off-by: Ziye Yang <ziye.yang@intel.com> Change-Id: I7b15c9c9a630ea7ecb2d3191b73c9c99f7febf31 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1189 Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
2020-03-09 15:26:21 +00:00
struct spdk_nvme_ctrlr ctrlr = {0};
unit/nvme: Add unit test for the tcp sgl It seems that autotest doesn't cover the function nvme_tcp_build_sgl_request(), So add the unit test. Change-Id: I2e243910cb38ffc49bdd6048b365dc7fa60892c2 Signed-off-by: JinYu <jin.yu@intel.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456728 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-06-04 10:47:09 +00:00
struct nvme_tcp_req tcp_req = {0};
struct nvme_request req = {{0}};
struct nvme_tcp_ut_bdev_io bio;
uint64_t i;
int rc;
lib/nvme_tcp: get the max_sges from the nvme ctrlr. Add the error print if there is still remaining_size in order to provide more meaningful debug info. Signed-off-by: Ziye Yang <ziye.yang@intel.com> Change-Id: I7b15c9c9a630ea7ecb2d3191b73c9c99f7febf31 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1189 Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
2020-03-09 15:26:21 +00:00
ctrlr.max_sges = NVME_TCP_MAX_SGL_DESCRIPTORS;
tqpair.qpair.ctrlr = &ctrlr;
unit/nvme: Add unit test for the tcp sgl It seems that autotest doesn't cover the function nvme_tcp_build_sgl_request(), So add the unit test. Change-Id: I2e243910cb38ffc49bdd6048b365dc7fa60892c2 Signed-off-by: JinYu <jin.yu@intel.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456728 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-06-04 10:47:09 +00:00
tcp_req.req = &req;
req.payload.reset_sgl_fn = nvme_tcp_ut_reset_sgl;
req.payload.next_sge_fn = nvme_tcp_ut_next_sge;
req.payload.contig_or_cb_arg = &bio;
req.qpair = &tqpair.qpair;
for (i = 0; i < NVME_TCP_MAX_SGL_DESCRIPTORS; i++) {
bio.iovs[i].iov_base = (void *)(0xFEEDB000 + i * 0x1000);
bio.iovs[i].iov_len = 0;
}
/* Test case 1: Single SGL. Expected: PASS */
bio.iovpos = 0;
req.payload_offset = 0;
req.payload_size = 0x1000;
bio.iovs[0].iov_len = 0x1000;
rc = nvme_tcp_build_sgl_request(&tqpair, &tcp_req);
SPDK_CU_ASSERT_FATAL(rc == 0);
CU_ASSERT(bio.iovpos == 1);
CU_ASSERT((uint64_t)tcp_req.iov[0].iov_base == (uint64_t)bio.iovs[0].iov_base);
CU_ASSERT(tcp_req.iov[0].iov_len == bio.iovs[0].iov_len);
CU_ASSERT(tcp_req.iovcnt == 1);
/* Test case 2: Multiple SGL. Expected: PASS */
bio.iovpos = 0;
req.payload_offset = 0;
req.payload_size = 0x4000;
for (i = 0; i < 4; i++) {
bio.iovs[i].iov_len = 0x1000;
}
rc = nvme_tcp_build_sgl_request(&tqpair, &tcp_req);
SPDK_CU_ASSERT_FATAL(rc == 0);
CU_ASSERT(bio.iovpos == 4);
CU_ASSERT(tcp_req.iovcnt == 4);
for (i = 0; i < 4; i++) {
CU_ASSERT(tcp_req.iov[i].iov_len == bio.iovs[i].iov_len);
CU_ASSERT((uint64_t)tcp_req.iov[i].iov_base == (uint64_t)bio.iovs[i].iov_base);
}
/* Test case 3: Payload is bigger than SGL. Expected: FAIL */
bio.iovpos = 0;
req.payload_offset = 0;
req.payload_size = 0x17000;
for (i = 0; i < NVME_TCP_MAX_SGL_DESCRIPTORS; i++) {
bio.iovs[i].iov_len = 0x1000;
}
rc = nvme_tcp_build_sgl_request(&tqpair, &tcp_req);
SPDK_CU_ASSERT_FATAL(rc != 0);
CU_ASSERT(bio.iovpos == NVME_TCP_MAX_SGL_DESCRIPTORS);
for (i = 0; i < NVME_TCP_MAX_SGL_DESCRIPTORS; i++) {
CU_ASSERT(tcp_req.iov[i].iov_len == bio.iovs[i].iov_len);
CU_ASSERT((uint64_t)tcp_req.iov[i].iov_base == (uint64_t)bio.iovs[i].iov_base);
}
}
nvme/tcp: Support unaligned extended LBA payload in nvme_tcp_pdu_set_data_buf When DIF is inserted or stripped, nvme_tcp_pdu_set_data_buf must shift the range of the buffer from LBA based to extended LBA based because the extended LBA based range must be passed to create a special temporary iovec array to leave a space of metadata for each block then. This patch do the following: - Add a pointer to the current DIF context to struct nvme_tcp_pdu. This pointer is set when DIF is inserted or stripped to the current NVMf request. - When the pointer to the current DIF context in the PDU is set, get the extended LBA based offset and length by calling spdk_dif_get_range_with_md() and use them in nvme_tcp_pdu_set_data_buf(). - Original data length is set in the PDU and original data offset is set in the current DIF context in nvme_tcp_pdu_set_data_buf(). - Add UT code. Change-Id: Ic1f4fd326b5b3e3d4c74677592bda34d6949a11e Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458920 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
2019-07-05 05:15:27 +00:00
static void
test_nvme_tcp_pdu_set_data_buf_with_md(void)
{
struct nvme_tcp_pdu pdu = {};
struct iovec iovs[7] = {};
struct spdk_dif_ctx dif_ctx = {};
int rc;
pdu.dif_ctx = &dif_ctx;
rc = spdk_dif_ctx_init(&dif_ctx, 520, 8, true, false, SPDK_DIF_DISABLE, 0,
0, 0, 0, 0, 0);
CU_ASSERT(rc == 0);
/* Single iovec case */
iovs[0].iov_base = (void *)0xDEADBEEF;
iovs[0].iov_len = 2080;
nvme_tcp_pdu_set_data_buf(&pdu, iovs, 1, 0, 500);
CU_ASSERT(dif_ctx.data_offset == 0);
CU_ASSERT(pdu.data_len == 500);
CU_ASSERT(pdu.data_iovcnt == 1);
CU_ASSERT(pdu.data_iov[0].iov_base == (void *)0xDEADBEEF);
CU_ASSERT(pdu.data_iov[0].iov_len == 500);
nvme_tcp_pdu_set_data_buf(&pdu, iovs, 1, 500, 1000);
CU_ASSERT(dif_ctx.data_offset == 500);
CU_ASSERT(pdu.data_len == 1000);
CU_ASSERT(pdu.data_iovcnt == 1);
CU_ASSERT(pdu.data_iov[0].iov_base == (void *)(0xDEADBEEF + 500));
CU_ASSERT(pdu.data_iov[0].iov_len == 1016);
nvme_tcp_pdu_set_data_buf(&pdu, iovs, 1, 1500, 548);
CU_ASSERT(dif_ctx.data_offset == 1500);
CU_ASSERT(pdu.data_len == 548);
CU_ASSERT(pdu.data_iovcnt == 1);
CU_ASSERT(pdu.data_iov[0].iov_base == (void *)(0xDEADBEEF + 1516));
CU_ASSERT(pdu.data_iov[0].iov_len == 564);
/* Multiple iovecs case */
iovs[0].iov_base = (void *)0xDEADBEEF;
iovs[0].iov_len = 256;
iovs[1].iov_base = (void *)((uint8_t *)(0xDEADBEEF + 0x1000));
iovs[1].iov_len = 256 + 1;
iovs[2].iov_base = (void *)((uint8_t *)(0xDEADBEEF + 0x2000));
iovs[2].iov_len = 4;
iovs[3].iov_base = (void *)((uint8_t *)(0xDEADBEEF + 0x3000));
iovs[3].iov_len = 3 + 123;
iovs[4].iov_base = (void *)((uint8_t *)(0xDEADBEEF + 0x4000));
iovs[4].iov_len = 389 + 6;
iovs[5].iov_base = (void *)((uint8_t *)(0xDEADBEEF + 0x5000));
iovs[5].iov_len = 2 + 512 + 8 + 432;
iovs[6].iov_base = (void *)((uint8_t *)(0xDEADBEEF + 0x6000));
iovs[6].iov_len = 80 + 8;
nvme_tcp_pdu_set_data_buf(&pdu, iovs, 7, 0, 500);
CU_ASSERT(dif_ctx.data_offset == 0);
CU_ASSERT(pdu.data_len == 500);
CU_ASSERT(pdu.data_iovcnt == 2);
CU_ASSERT(pdu.data_iov[0].iov_base == (void *)0xDEADBEEF);
CU_ASSERT(pdu.data_iov[0].iov_len == 256);
CU_ASSERT(pdu.data_iov[1].iov_base == (void *)(0xDEADBEEF + 0x1000));
CU_ASSERT(pdu.data_iov[1].iov_len == 244);
nvme_tcp_pdu_set_data_buf(&pdu, iovs, 7, 500, 1000);
CU_ASSERT(dif_ctx.data_offset == 500);
CU_ASSERT(pdu.data_len == 1000);
CU_ASSERT(pdu.data_iovcnt == 5);
CU_ASSERT(pdu.data_iov[0].iov_base == (void *)(0xDEADBEEF + 0x1000 + 244));
CU_ASSERT(pdu.data_iov[0].iov_len == 13);
CU_ASSERT(pdu.data_iov[1].iov_base == (void *)(0xDEADBEEF + 0x2000));
CU_ASSERT(pdu.data_iov[1].iov_len == 4);
CU_ASSERT(pdu.data_iov[2].iov_base == (void *)(0xDEADBEEF + 0x3000));
CU_ASSERT(pdu.data_iov[2].iov_len == 3 + 123);
CU_ASSERT(pdu.data_iov[3].iov_base == (void *)(0xDEADBEEF + 0x4000));
CU_ASSERT(pdu.data_iov[3].iov_len == 395);
CU_ASSERT(pdu.data_iov[4].iov_base == (void *)(0xDEADBEEF + 0x5000));
CU_ASSERT(pdu.data_iov[4].iov_len == 478);
nvme_tcp_pdu_set_data_buf(&pdu, iovs, 7, 1500, 548);
CU_ASSERT(dif_ctx.data_offset == 1500);
CU_ASSERT(pdu.data_len == 548);
CU_ASSERT(pdu.data_iovcnt == 2);
CU_ASSERT(pdu.data_iov[0].iov_base == (void *)(0xDEADBEEF + 0x5000 + 478));
CU_ASSERT(pdu.data_iov[0].iov_len == 476);
CU_ASSERT(pdu.data_iov[1].iov_base == (void *)(0xDEADBEEF + 0x6000));
CU_ASSERT(pdu.data_iov[1].iov_len == 88);
}
nvme/tcp: Support extended LBA payload in nvme_tcp_build_iovs When pdu->dif_ctx is not NULL, call spdk_dif_set_md_interleave_iovs in nvme_tcp_build_iovs and nvme_tcp_build_payload_iovs to create special SGL for the data segment to leave a space of metadata for each block. Add UT to verify if the update works correctly. This update is very similar with the use case of spdk_dif_set_md_interleave_iovs in iSCSI target. Change-Id: Ie802cb4b7c541089579ecc2a630fc1e34004da55 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458923 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-07-03 04:53:36 +00:00
static void
test_nvme_tcp_build_iovs_with_md(void)
{
struct nvme_tcp_pdu pdu = {};
struct iovec iovs[11] = {};
struct spdk_dif_ctx dif_ctx = {};
uint32_t mapped_length = 0;
int rc;
rc = spdk_dif_ctx_init(&dif_ctx, 520, 8, true, false, SPDK_DIF_DISABLE, 0,
0, 0, 0, 0, 0);
CU_ASSERT(rc == 0);
pdu.dif_ctx = &dif_ctx;
Revert "nvme/tcp: Change hdr in nvme_tcp_pdu to pointer" This reverts commit ea5ad0b28630bf735b3efbb126c0ac00b0403647. This code is moving from the nvmf target to the posix sock layer in this series. Change-Id: I333bdf325848e726ab82a9e6916e1bbdcd34009c Signed-off-by: Ben Walker <benjamin.walker@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/446 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2019-09-27 21:19:57 +00:00
pdu.hdr.common.pdu_type = SPDK_NVME_TCP_PDU_TYPE_CAPSULE_CMD;
pdu.hdr.common.hlen = sizeof(struct spdk_nvme_tcp_cmd);
pdu.hdr.common.plen = pdu.hdr.common.hlen + SPDK_NVME_TCP_DIGEST_LEN + 512 * 8 +
SPDK_NVME_TCP_DIGEST_LEN;
nvme/tcp: Support extended LBA payload in nvme_tcp_build_iovs When pdu->dif_ctx is not NULL, call spdk_dif_set_md_interleave_iovs in nvme_tcp_build_iovs and nvme_tcp_build_payload_iovs to create special SGL for the data segment to leave a space of metadata for each block. Add UT to verify if the update works correctly. This update is very similar with the use case of spdk_dif_set_md_interleave_iovs in iSCSI target. Change-Id: Ie802cb4b7c541089579ecc2a630fc1e34004da55 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458923 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-07-03 04:53:36 +00:00
pdu.data_len = 512 * 8;
pdu.padding_len = 0;
pdu.data_iov[0].iov_base = (void *)0xDEADBEEF;
pdu.data_iov[0].iov_len = (512 + 8) * 8;
pdu.data_iovcnt = 1;
rc = nvme_tcp_build_iovs(iovs, 11, &pdu, true, true, &mapped_length);
CU_ASSERT(rc == 10);
Revert "nvme/tcp: Change hdr in nvme_tcp_pdu to pointer" This reverts commit ea5ad0b28630bf735b3efbb126c0ac00b0403647. This code is moving from the nvmf target to the posix sock layer in this series. Change-Id: I333bdf325848e726ab82a9e6916e1bbdcd34009c Signed-off-by: Ben Walker <benjamin.walker@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/446 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Jim Harris <james.r.harris@intel.com>
2019-09-27 21:19:57 +00:00
CU_ASSERT(iovs[0].iov_base == (void *)&pdu.hdr.raw);
nvme/tcp: Support extended LBA payload in nvme_tcp_build_iovs When pdu->dif_ctx is not NULL, call spdk_dif_set_md_interleave_iovs in nvme_tcp_build_iovs and nvme_tcp_build_payload_iovs to create special SGL for the data segment to leave a space of metadata for each block. Add UT to verify if the update works correctly. This update is very similar with the use case of spdk_dif_set_md_interleave_iovs in iSCSI target. Change-Id: Ie802cb4b7c541089579ecc2a630fc1e34004da55 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/458923 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-07-03 04:53:36 +00:00
CU_ASSERT(iovs[0].iov_len == sizeof(struct spdk_nvme_tcp_cmd) + SPDK_NVME_TCP_DIGEST_LEN);
CU_ASSERT(iovs[1].iov_base == (void *)0xDEADBEEF);
CU_ASSERT(iovs[1].iov_len == 512);
CU_ASSERT(iovs[2].iov_base == (void *)(0xDEADBEEF + 520));
CU_ASSERT(iovs[2].iov_len == 512);
CU_ASSERT(iovs[3].iov_base == (void *)(0xDEADBEEF + 520 * 2));
CU_ASSERT(iovs[3].iov_len == 512);
CU_ASSERT(iovs[4].iov_base == (void *)(0xDEADBEEF + 520 * 3));
CU_ASSERT(iovs[4].iov_len == 512);
CU_ASSERT(iovs[5].iov_base == (void *)(0xDEADBEEF + 520 * 4));
CU_ASSERT(iovs[5].iov_len == 512);
CU_ASSERT(iovs[6].iov_base == (void *)(0xDEADBEEF + 520 * 5));
CU_ASSERT(iovs[6].iov_len == 512);
CU_ASSERT(iovs[7].iov_base == (void *)(0xDEADBEEF + 520 * 6));
CU_ASSERT(iovs[7].iov_len == 512);
CU_ASSERT(iovs[8].iov_base == (void *)(0xDEADBEEF + 520 * 7));
CU_ASSERT(iovs[8].iov_len == 512);
CU_ASSERT(iovs[9].iov_base == (void *)pdu.data_digest);
CU_ASSERT(iovs[9].iov_len == SPDK_NVME_TCP_DIGEST_LEN);
CU_ASSERT(mapped_length == sizeof(struct spdk_nvme_tcp_cmd) + SPDK_NVME_TCP_DIGEST_LEN +
512 * 8 + SPDK_NVME_TCP_DIGEST_LEN);
}
test/nvme_tcp: improve unittest coverage of tcp Cases for nvme tcp requests compelted safety. Change-Id: Iff5d2418881de8aa1e29fa38149744740772be11 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5856 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Mellanox Build Bot Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
2021-01-05 08:09:09 +00:00
/* Just define, nothing to do */
static void
ut_nvme_complete_request(void *arg, const struct spdk_nvme_cpl *cpl)
{
return;
}
static void
test_nvme_tcp_req_complete_safe(void)
{
bool rc;
ut/nvme: Fix compilation warning GCC 4.8.5 compains on missing braces: nvme_tcp_ut.c: In function 'test_nvme_tcp_req_complete_safe': nvme_tcp_ut.c:456:9: warning: missing braces around initializer [-Wmissing-braces] struct nvme_request req = {0}; ^ nvme_tcp_ut.c:456:9: warning: (near initialization for 'req.cmd') [-Wmissing-braces] nvme_tcp_ut.c:458:9: warning: missing braces around initializer [-Wmissing-braces] struct nvme_tcp_qpair tqpair = {0}; ^ nvme_tcp_ut.c:458:9: warning: (near initialization for 'tqpair.qpair') [-Wmissing-braces] Also structure spdk_nvme_qpair is embedded into nvme_tcp_qpair, there is no need in a separate variable. Change-Id: I3a9d6760a0cffe90ece0a6d470dda8309a6619e7 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5985 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: <dongx.yi@intel.com> Community-CI: Mellanox Build Bot
2021-01-19 15:22:02 +00:00
struct nvme_tcp_req tcp_req = {0};
struct nvme_request req = {{0}};
struct nvme_tcp_qpair tqpair = {{0}};
test/nvme_tcp: improve unittest coverage of tcp Cases for nvme tcp requests compelted safety. Change-Id: Iff5d2418881de8aa1e29fa38149744740772be11 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5856 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Mellanox Build Bot Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
2021-01-05 08:09:09 +00:00
tcp_req.req = &req;
ut/nvme: Fix compilation warning GCC 4.8.5 compains on missing braces: nvme_tcp_ut.c: In function 'test_nvme_tcp_req_complete_safe': nvme_tcp_ut.c:456:9: warning: missing braces around initializer [-Wmissing-braces] struct nvme_request req = {0}; ^ nvme_tcp_ut.c:456:9: warning: (near initialization for 'req.cmd') [-Wmissing-braces] nvme_tcp_ut.c:458:9: warning: missing braces around initializer [-Wmissing-braces] struct nvme_tcp_qpair tqpair = {0}; ^ nvme_tcp_ut.c:458:9: warning: (near initialization for 'tqpair.qpair') [-Wmissing-braces] Also structure spdk_nvme_qpair is embedded into nvme_tcp_qpair, there is no need in a separate variable. Change-Id: I3a9d6760a0cffe90ece0a6d470dda8309a6619e7 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5985 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: <dongx.yi@intel.com> Community-CI: Mellanox Build Bot
2021-01-19 15:22:02 +00:00
tcp_req.req->qpair = &tqpair.qpair;
test/nvme_tcp: improve unittest coverage of tcp Cases for nvme tcp requests compelted safety. Change-Id: Iff5d2418881de8aa1e29fa38149744740772be11 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5856 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Mellanox Build Bot Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
2021-01-05 08:09:09 +00:00
tcp_req.req->cb_fn = ut_nvme_complete_request;
tcp_req.tqpair = &tqpair;
tcp_req.state = NVME_TCP_REQ_ACTIVE;
TAILQ_INIT(&tcp_req.tqpair->outstanding_reqs);
/* Test case 1: send operation and transfer completed. Expect: PASS */
tcp_req.state = NVME_TCP_REQ_ACTIVE;
tcp_req.ordering.bits.send_ack = 1;
tcp_req.ordering.bits.data_recv = 1;
TAILQ_INSERT_TAIL(&tcp_req.tqpair->outstanding_reqs, &tcp_req, link);
rc = nvme_tcp_req_complete_safe(&tcp_req);
CU_ASSERT(rc == true);
/* Test case 2: send operation not completed. Expect: FAIL */
tcp_req.ordering.raw = 0;
tcp_req.state = NVME_TCP_REQ_ACTIVE;
TAILQ_INSERT_TAIL(&tcp_req.tqpair->outstanding_reqs, &tcp_req, link);
rc = nvme_tcp_req_complete_safe(&tcp_req);
SPDK_CU_ASSERT_FATAL(rc != true);
TAILQ_REMOVE(&tcp_req.tqpair->outstanding_reqs, &tcp_req, link);
/* Test case 3: in completion context. Expect: PASS */
tqpair.qpair.in_completion_context = 1;
tqpair.async_complete = 0;
tcp_req.ordering.bits.send_ack = 1;
tcp_req.ordering.bits.data_recv = 1;
tcp_req.state = NVME_TCP_REQ_ACTIVE;
TAILQ_INSERT_TAIL(&tcp_req.tqpair->outstanding_reqs, &tcp_req, link);
rc = nvme_tcp_req_complete_safe(&tcp_req);
CU_ASSERT(rc == true);
CU_ASSERT(tcp_req.tqpair->async_complete == 0);
/* Test case 4: in async complete. Expect: PASS */
tqpair.qpair.in_completion_context = 0;
tcp_req.ordering.bits.send_ack = 1;
tcp_req.ordering.bits.data_recv = 1;
tcp_req.state = NVME_TCP_REQ_ACTIVE;
TAILQ_INSERT_TAIL(&tcp_req.tqpair->outstanding_reqs, &tcp_req, link);
rc = nvme_tcp_req_complete_safe(&tcp_req);
CU_ASSERT(rc == true);
CU_ASSERT(tcp_req.tqpair->async_complete);
}
test/nvme_tcp: nvme tcp UT cases Cases for getting tcp requests and request initialization. Change-Id: I64ba6dbfba6bd68b70a3b7f1090ea274e3195634 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6140 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Community-CI: Mellanox Build Bot
2021-01-28 06:45:42 +00:00
static void
test_nvme_tcp_req_init(void)
{
struct nvme_tcp_qpair tqpair = {0};
struct nvme_request req = {0};
struct nvme_tcp_req tcp_req = {0};
struct spdk_nvme_ctrlr ctrlr = {0};
struct nvme_tcp_ut_bdev_io bio = {0};
int rc;
tqpair.qpair.ctrlr = &ctrlr;
req.qpair = &tqpair.qpair;
tcp_req.cid = 1;
req.payload.next_sge_fn = nvme_tcp_ut_next_sge;
req.payload.contig_or_cb_arg = &bio;
req.payload_offset = 0;
req.payload_size = 4096;
ctrlr.max_sges = NVME_TCP_MAX_SGL_DESCRIPTORS;
ctrlr.ioccsz_bytes = 1024;
bio.iovpos = 0;
bio.iovs[0].iov_len = 8192;
bio.iovs[0].iov_base = (void *)0xDEADBEEF;
/* Test case1: payload type SGL. Expect: PASS */
req.cmd.opc = SPDK_NVME_DATA_HOST_TO_CONTROLLER;
req.payload.reset_sgl_fn = nvme_tcp_ut_reset_sgl;
rc = nvme_tcp_req_init(&tqpair, &req, &tcp_req);
CU_ASSERT(rc == 0);
CU_ASSERT(tcp_req.req == &req);
CU_ASSERT(tcp_req.in_capsule_data == true);
CU_ASSERT(tcp_req.iovcnt == 1);
CU_ASSERT(tcp_req.iov[0].iov_len == req.payload_size);
CU_ASSERT(tcp_req.iov[0].iov_base == bio.iovs[0].iov_base);
CU_ASSERT(req.cmd.cid == tcp_req.cid);
CU_ASSERT(req.cmd.psdt == SPDK_NVME_PSDT_SGL_MPTR_CONTIG);
CU_ASSERT(req.cmd.dptr.sgl1.unkeyed.type == SPDK_NVME_SGL_TYPE_DATA_BLOCK);
CU_ASSERT(req.cmd.dptr.sgl1.unkeyed.subtype == SPDK_NVME_SGL_SUBTYPE_OFFSET);
CU_ASSERT(req.cmd.dptr.sgl1.unkeyed.length == req.payload_size);
CU_ASSERT(req.cmd.dptr.sgl1.address == 0);
/* Test case2: payload type CONTIG. Expect: PASS */
memset(&req.cmd, 0, sizeof(req.cmd));
memset(&tcp_req, 0, sizeof(tcp_req));
tcp_req.cid = 1;
req.payload.reset_sgl_fn = NULL;
req.cmd.opc = SPDK_NVME_DATA_HOST_TO_CONTROLLER;
rc = nvme_tcp_req_init(&tqpair, &req, &tcp_req);
CU_ASSERT(rc == 0);
CU_ASSERT(tcp_req.req == &req);
CU_ASSERT(tcp_req.in_capsule_data == true);
CU_ASSERT(tcp_req.iov[0].iov_len == req.payload_size);
CU_ASSERT(tcp_req.iov[0].iov_base == &bio);
CU_ASSERT(tcp_req.iovcnt == 1);
CU_ASSERT(req.cmd.cid == tcp_req.cid);
CU_ASSERT(req.cmd.psdt == SPDK_NVME_PSDT_SGL_MPTR_CONTIG);
CU_ASSERT(req.cmd.dptr.sgl1.unkeyed.type == SPDK_NVME_SGL_TYPE_DATA_BLOCK);
CU_ASSERT(req.cmd.dptr.sgl1.unkeyed.subtype == SPDK_NVME_SGL_SUBTYPE_OFFSET);
CU_ASSERT(req.cmd.dptr.sgl1.unkeyed.length == req.payload_size);
CU_ASSERT(req.cmd.dptr.sgl1.address == 0);
}
static void
test_nvme_tcp_req_get(void)
{
struct nvme_tcp_req tcp_req = {0};
struct nvme_tcp_qpair tqpair = {0};
struct nvme_tcp_pdu send_pdu = {0};
tcp_req.send_pdu = &send_pdu;
tcp_req.state = NVME_TCP_REQ_FREE;
TAILQ_INIT(&tqpair.free_reqs);
TAILQ_INIT(&tqpair.outstanding_reqs);
TAILQ_INSERT_HEAD(&tqpair.free_reqs, &tcp_req, link);
CU_ASSERT(nvme_tcp_req_get(&tqpair) == &tcp_req);
CU_ASSERT(tcp_req.state == NVME_TCP_REQ_ACTIVE);
CU_ASSERT(tcp_req.datao == 0);
CU_ASSERT(tcp_req.req == NULL);
CU_ASSERT(tcp_req.in_capsule_data == false);
CU_ASSERT(tcp_req.r2tl_remain == 0);
CU_ASSERT(tcp_req.iovcnt == 0);
CU_ASSERT(tcp_req.ordering.raw == 0);
CU_ASSERT(!TAILQ_EMPTY(&tqpair.outstanding_reqs));
CU_ASSERT(TAILQ_EMPTY(&tqpair.free_reqs));
/* No tcp request available, expect fail */
SPDK_CU_ASSERT_FATAL(nvme_tcp_req_get(&tqpair) == NULL);
}
test/nvme_tcp: Cases for nvme tcp Cases for sending tcp qpair command and writing tcp qpair pdu. Change-Id: I183d61fefd0749fb8a269d4f240d6e2e917a4996 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5998 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Mellanox Build Bot Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
2021-02-19 02:35:39 +00:00
static void
test_nvme_tcp_qpair_capsule_cmd_send(void)
{
struct nvme_tcp_qpair tqpair = {};
struct nvme_tcp_req tcp_req = {};
struct nvme_tcp_pdu pdu = {};
struct nvme_request req = {};
char iov_base0[4096];
char iov_base1[4096];
uint32_t plen;
uint8_t pdo;
memset(iov_base0, 0xFF, 4096);
memset(iov_base1, 0xFF, 4096);
tcp_req.req = &req;
tcp_req.send_pdu = &pdu;
TAILQ_INIT(&tqpair.send_queue);
tcp_req.iov[0].iov_base = (void *)iov_base0;
tcp_req.iov[0].iov_len = 4096;
tcp_req.iov[1].iov_base = (void *)iov_base1;
tcp_req.iov[1].iov_len = 4096;
tcp_req.iovcnt = 2;
tcp_req.req->payload_size = 8192;
tcp_req.in_capsule_data = true;
tqpair.cpda = NVME_TCP_HPDA_DEFAULT;
/* Test case 1: host hdgst and ddgst enable. Expect: PASS */
tqpair.flags.host_hdgst_enable = 1;
tqpair.flags.host_ddgst_enable = 1;
pdo = plen = sizeof(struct spdk_nvme_tcp_cmd) +
SPDK_NVME_TCP_DIGEST_LEN;
plen += tcp_req.req->payload_size;
plen += SPDK_NVME_TCP_DIGEST_LEN;
nvme_tcp_qpair_capsule_cmd_send(&tqpair, &tcp_req);
TAILQ_REMOVE(&tqpair.send_queue, &pdu, tailq);
CU_ASSERT(pdu.hdr.capsule_cmd.common.flags
& SPDK_NVME_TCP_CH_FLAGS_HDGSTF);
CU_ASSERT(pdu.hdr.capsule_cmd.common.flags
& SPDK_NVME_TCP_CH_FLAGS_DDGSTF);
CU_ASSERT(pdu.hdr.capsule_cmd.common.pdu_type ==
SPDK_NVME_TCP_PDU_TYPE_CAPSULE_CMD);
CU_ASSERT(pdu.hdr.capsule_cmd.common.pdo == pdo);
CU_ASSERT(pdu.hdr.capsule_cmd.common.plen == plen);
CU_ASSERT(pdu.data_iov[0].iov_base == tcp_req.iov[0].iov_base);
CU_ASSERT(pdu.data_iov[0].iov_len == tcp_req.iov[0].iov_len);
CU_ASSERT(pdu.data_iov[1].iov_base == tcp_req.iov[1].iov_base);
CU_ASSERT(pdu.data_iov[1].iov_len == tcp_req.iov[0].iov_len);
/* Test case 2: host hdgst and ddgst disable. Expect: PASS */
memset(&pdu, 0, sizeof(pdu));
tqpair.flags.host_hdgst_enable = 0;
tqpair.flags.host_ddgst_enable = 0;
pdo = plen = sizeof(struct spdk_nvme_tcp_cmd);
plen += tcp_req.req->payload_size;
nvme_tcp_qpair_capsule_cmd_send(&tqpair, &tcp_req);
TAILQ_REMOVE(&tqpair.send_queue, &pdu, tailq);
CU_ASSERT(pdu.hdr.capsule_cmd.common.flags == 0)
CU_ASSERT(pdu.hdr.capsule_cmd.common.pdu_type ==
SPDK_NVME_TCP_PDU_TYPE_CAPSULE_CMD);
CU_ASSERT(pdu.hdr.capsule_cmd.common.pdo == pdo);
CU_ASSERT(pdu.hdr.capsule_cmd.common.plen == plen);
CU_ASSERT(pdu.data_iov[0].iov_base == tcp_req.iov[0].iov_base);
CU_ASSERT(pdu.data_iov[0].iov_len == tcp_req.iov[0].iov_len);
CU_ASSERT(pdu.data_iov[1].iov_base == tcp_req.iov[1].iov_base);
CU_ASSERT(pdu.data_iov[1].iov_len == tcp_req.iov[0].iov_len);
/* Test case 3: padding available. Expect: PASS */
memset(&pdu, 0, sizeof(pdu));
tqpair.flags.host_hdgst_enable = 1;
tqpair.flags.host_ddgst_enable = 1;
tqpair.cpda = SPDK_NVME_TCP_CPDA_MAX;
pdo = plen = (SPDK_NVME_TCP_CPDA_MAX + 1) << 2;
plen += tcp_req.req->payload_size;
plen += SPDK_NVME_TCP_DIGEST_LEN;
nvme_tcp_qpair_capsule_cmd_send(&tqpair, &tcp_req);
TAILQ_REMOVE(&tqpair.send_queue, &pdu, tailq);
CU_ASSERT(pdu.hdr.capsule_cmd.common.flags
& SPDK_NVME_TCP_CH_FLAGS_HDGSTF);
CU_ASSERT(pdu.hdr.capsule_cmd.common.flags
& SPDK_NVME_TCP_CH_FLAGS_DDGSTF);
CU_ASSERT(pdu.hdr.capsule_cmd.common.pdu_type ==
SPDK_NVME_TCP_PDU_TYPE_CAPSULE_CMD);
CU_ASSERT(pdu.hdr.capsule_cmd.common.pdo == pdo);
CU_ASSERT(pdu.hdr.capsule_cmd.common.plen == plen);
CU_ASSERT(pdu.data_iov[0].iov_base == tcp_req.iov[0].iov_base);
CU_ASSERT(pdu.data_iov[0].iov_len == tcp_req.iov[0].iov_len);
CU_ASSERT(pdu.data_iov[1].iov_base == tcp_req.iov[1].iov_base);
CU_ASSERT(pdu.data_iov[1].iov_len == tcp_req.iov[0].iov_len);
}
/* Just define, nothing to do */
static void
ut_nvme_tcp_qpair_xfer_complete_cb(void *cb_arg)
{
return;
}
static void
test_nvme_tcp_qpair_write_pdu(void)
{
struct nvme_tcp_qpair tqpair = {};
struct nvme_tcp_pdu pdu = {};
void *cb_arg = (void *)0xDEADBEEF;
char iov_base0[4096];
char iov_base1[4096];
memset(iov_base0, 0xFF, 4096);
memset(iov_base1, 0xFF, 4096);
pdu.data_len = 4096 * 2;
pdu.padding_len = 0;
pdu.data_iov[0].iov_base = (void *)iov_base0;
pdu.data_iov[0].iov_len = 4096;
pdu.data_iov[1].iov_base = (void *)iov_base1;
pdu.data_iov[1].iov_len = 4096;
pdu.data_iovcnt = 2;
TAILQ_INIT(&tqpair.send_queue);
/* Test case1: host hdgst and ddgst enable Expect: PASS */
memset(pdu.hdr.raw, 0, SPDK_NVME_TCP_TERM_REQ_PDU_MAX_SIZE);
memset(pdu.data_digest, 0, SPDK_NVME_TCP_DIGEST_LEN);
pdu.hdr.common.pdu_type = SPDK_NVME_TCP_PDU_TYPE_CAPSULE_CMD;
pdu.hdr.common.hlen = sizeof(struct spdk_nvme_tcp_cmd);
pdu.hdr.common.plen = pdu.hdr.common.hlen +
SPDK_NVME_TCP_DIGEST_LEN * 2 ;
pdu.hdr.common.plen += pdu.data_len;
tqpair.flags.host_hdgst_enable = 1;
tqpair.flags.host_ddgst_enable = 1;
nvme_tcp_qpair_write_pdu(&tqpair,
&pdu,
ut_nvme_tcp_qpair_xfer_complete_cb,
cb_arg);
TAILQ_REMOVE(&tqpair.send_queue, &pdu, tailq);
/* Check the crc data of header digest filled into raw */
CU_ASSERT(pdu.hdr.raw[pdu.hdr.common.hlen]);
CU_ASSERT(pdu.data_digest[0]);
CU_ASSERT(pdu.sock_req.iovcnt == 4);
CU_ASSERT(pdu.iov[0].iov_base == &pdu.hdr.raw);
CU_ASSERT(pdu.iov[0].iov_len == (sizeof(struct spdk_nvme_tcp_cmd) +
SPDK_NVME_TCP_DIGEST_LEN));
CU_ASSERT(pdu.iov[1].iov_base == pdu.data_iov[0].iov_base);
CU_ASSERT(pdu.iov[1].iov_len == pdu.data_iov[0].iov_len);
CU_ASSERT(pdu.iov[2].iov_base == pdu.data_iov[1].iov_base);
CU_ASSERT(pdu.iov[2].iov_len == pdu.data_iov[1].iov_len);
CU_ASSERT(pdu.iov[3].iov_base == &pdu.data_digest);
CU_ASSERT(pdu.iov[3].iov_len == SPDK_NVME_TCP_DIGEST_LEN);
CU_ASSERT(pdu.cb_fn == ut_nvme_tcp_qpair_xfer_complete_cb);
CU_ASSERT(pdu.cb_arg == cb_arg);
CU_ASSERT(pdu.qpair == &tqpair);
CU_ASSERT(pdu.sock_req.cb_arg == (void *)&pdu);
/* Test case2: host hdgst and ddgst disable Expect: PASS */
memset(pdu.hdr.raw, 0, SPDK_NVME_TCP_TERM_REQ_PDU_MAX_SIZE);
memset(pdu.data_digest, 0, SPDK_NVME_TCP_DIGEST_LEN);
pdu.hdr.common.hlen = sizeof(struct spdk_nvme_tcp_cmd);
pdu.hdr.common.plen = pdu.hdr.common.hlen + pdu.data_len;
tqpair.flags.host_hdgst_enable = 0;
tqpair.flags.host_ddgst_enable = 0;
nvme_tcp_qpair_write_pdu(&tqpair,
&pdu,
ut_nvme_tcp_qpair_xfer_complete_cb,
cb_arg);
TAILQ_REMOVE(&tqpair.send_queue, &pdu, tailq);
CU_ASSERT(pdu.hdr.raw[pdu.hdr.common.hlen] == 0);
CU_ASSERT(pdu.data_digest[0] == 0);
CU_ASSERT(pdu.sock_req.iovcnt == 3);
CU_ASSERT(pdu.iov[0].iov_base == &pdu.hdr.raw);
CU_ASSERT(pdu.iov[0].iov_len == sizeof(struct spdk_nvme_tcp_cmd));
CU_ASSERT(pdu.iov[1].iov_base == pdu.data_iov[0].iov_base);
CU_ASSERT(pdu.iov[1].iov_len == pdu.data_iov[0].iov_len);
CU_ASSERT(pdu.iov[2].iov_base == pdu.data_iov[1].iov_base);
CU_ASSERT(pdu.iov[2].iov_len == pdu.data_iov[1].iov_len);
CU_ASSERT(pdu.cb_fn == ut_nvme_tcp_qpair_xfer_complete_cb);
CU_ASSERT(pdu.cb_arg == cb_arg);
CU_ASSERT(pdu.qpair == &tqpair);
CU_ASSERT(pdu.sock_req.cb_arg == (void *)&pdu);
}
nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
int main(int argc, char **argv)
{
CU_pSuite suite = NULL;
unsigned int num_failures;
ut: simplify error handling at initialization This patch changes the way that unittests initialize. The new way is shorter and simpler. It assumes that error during initialization is a fatal error, but in our cases it always is. This patch is a followp up to https://review.spdk.io/gerrit/c/spdk/spdk/+/930 lvol_ut.c is the only test that skipped because it runs same tests multiple times which is not allowed by new method. Change-Id: I0baf7ad09a35d5fca2dc4a03b4453c12c14f38a7 Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1238 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
2020-03-11 17:59:24 +00:00
CU_set_error_action(CUEA_ABORT);
CU_initialize_registry();
nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
suite = CU_add_suite("nvme_tcp", NULL, NULL);
ut: make use of CUnit macro to reduce duplications CUnit provides a helper macro CU_ADD_TEST() that simplifies usage of CU_add_test() function. Test name no longer needs to be provided, as it will be inherited from function name. This is a follow up to https://review.spdk.io/gerrit/c/spdk/spdk/+/931 Change-Id: I8078f02e08b14f12328ae022d7090ba13fbd64e4 Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1239 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
2020-03-11 19:15:39 +00:00
CU_ADD_TEST(suite, test_nvme_tcp_pdu_set_data_buf);
CU_ADD_TEST(suite, test_nvme_tcp_build_iovs);
CU_ADD_TEST(suite, test_nvme_tcp_build_sgl_request);
CU_ADD_TEST(suite, test_nvme_tcp_pdu_set_data_buf_with_md);
CU_ADD_TEST(suite, test_nvme_tcp_build_iovs_with_md);
test/nvme_tcp: improve unittest coverage of tcp Cases for nvme tcp requests compelted safety. Change-Id: Iff5d2418881de8aa1e29fa38149744740772be11 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5856 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Mellanox Build Bot Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
2021-01-05 08:09:09 +00:00
CU_ADD_TEST(suite, test_nvme_tcp_req_complete_safe);
test/nvme_tcp: nvme tcp UT cases Cases for getting tcp requests and request initialization. Change-Id: I64ba6dbfba6bd68b70a3b7f1090ea274e3195634 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6140 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Community-CI: Mellanox Build Bot
2021-01-28 06:45:42 +00:00
CU_ADD_TEST(suite, test_nvme_tcp_req_get);
CU_ADD_TEST(suite, test_nvme_tcp_req_init);
test/nvme_tcp: Cases for nvme tcp Cases for sending tcp qpair command and writing tcp qpair pdu. Change-Id: I183d61fefd0749fb8a269d4f240d6e2e917a4996 Signed-off-by: Mao Jiang <maox.jiang@intel.com> Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5998 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Mellanox Build Bot Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
2021-02-19 02:35:39 +00:00
CU_ADD_TEST(suite, test_nvme_tcp_qpair_capsule_cmd_send);
CU_ADD_TEST(suite, test_nvme_tcp_qpair_write_pdu);
nvme/tcp: add unit test to verify the bug in nvmf_tcp_pdu_set_data_buf This patch adds UT code to test if the bug in nvmf_tcp_pdu_set_data_buf is fixed by the last patch. Change-Id: I0b504ba37652de8300c85eea2f85218c837dd503 Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455624 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
2019-05-24 04:01:05 +00:00
CU_basic_set_mode(CU_BRM_VERBOSE);
CU_basic_run_tests();
num_failures = CU_get_number_of_failures();
CU_cleanup_registry();
return num_failures;
}
Reference in New Issue Copy Permalink