On 7/21/2023 6:42 PM, Ivan Malov wrote:
> Hi Ferruh,
>
> PSB
>
> Thank you.
>
> On Fri, 21 Jul 2023, Ferruh Yigit wrote:
>
>> Drivers start/stop device queues on port start/stop, but not all drivers
>> update queue state accordingly.
>>
>> This becomes more visible if a specific queue stopped explicitly and
>> port stopped/started later, in this case although all queues are
>> started, the state of that specific queue is stopped and it is
>> misleading.
>>
>> Misrepresentation of queue state became a defect with commit [1] that
>> does forwarding decision based on queue state and commit [2] that gets
>> up to date queue state from ethdev/device before forwarding.
>>
>> This patch documents that status of all queues of a device should be
>> `RTE_ETH_QUEUE_STATE_STOPPED` after port stop and their status should
>> be`RTE_ETH_QUEUE_STATE_STARTED` after port start.
>>
>> Also an unit test added to verify drivers.
>>
>> Signed-off-by: Ferruh Yigit <ferruh.yi...@amd.com>
>> ---
>> Cc: Jie Hai <haij...@huawei.com>
>> Cc: Song Jiale <songx.ji...@intel.com>
>> Cc: Yuan Peng <yuan.p...@intel.com>
>> Cc: Raslan Darawsheh <rasl...@nvidia.com>
>> Cc: Qiming Yang <qiming.y...@intel.com>
>> ---
>> app/test/meson.build | 2 +
>> app/test/test_ethdev_api.c | 169 +++++++++++++++++++++++++++++++++++++
>> lib/ethdev/rte_ethdev.h | 5 ++
>> 3 files changed, 176 insertions(+)
>> create mode 100644 app/test/test_ethdev_api.c
>>
>> diff --git a/app/test/meson.build b/app/test/meson.build
>> index b89cf0368fb5..8e409cf59c35 100644
>> --- a/app/test/meson.build
>> +++ b/app/test/meson.build
>> @@ -49,6 +49,7 @@ test_sources = files(
>> 'test_efd_perf.c',
>> 'test_errno.c',
>> 'test_ethdev_link.c',
>> + 'test_ethdev_api.c',
>> 'test_event_crypto_adapter.c',
>> 'test_event_eth_rx_adapter.c',
>> 'test_event_ring.c',
>> @@ -187,6 +188,7 @@ fast_tests = [
>> ['eal_fs_autotest', true, true],
>> ['errno_autotest', true, true],
>> ['ethdev_link_status', true, true],
>> + ['ethdev_api', true, true],
>> ['event_ring_autotest', true, true],
>> ['fib_autotest', true, true],
>> ['fib6_autotest', true, true],
>> diff --git a/app/test/test_ethdev_api.c b/app/test/test_ethdev_api.c
>> new file mode 100644
>> index 000000000000..1b4569396dda
>> --- /dev/null
>> +++ b/app/test/test_ethdev_api.c
>> @@ -0,0 +1,169 @@
>> +/* SPDX-License-Identifier: BSD-3-Clause
>> + * Copyright (C) 2023, Advanced Micro Devices, Inc.
>> + */
>> +
>> +#include <rte_log.h>
>> +#include <rte_ethdev.h>
>> +
>> +#include <rte_test.h>
>> +#include "test.h"
>> +
>> +#define NUM_RXQ 2
>> +#define NUM_TXQ 2
>> +#define NUM_RXD 512
>> +#define NUM_TXD 512
>> +#define NUM_MBUF 1024
>> +#define MBUF_CACHE_SIZE 256
>> +
>> +static int32_t
>> +ethdev_api_queue_status(void)
>> +{
>> + struct rte_eth_dev_info dev_info;
>> + struct rte_eth_rxq_info rx_qinfo;
>> + struct rte_eth_txq_info tx_qinfo;
>> + struct rte_mempool *mbuf_pool;
>> + /*struct rte_eth_rxconf rx_conf;*/
>> + /*struct rte_eth_txconf tx_conf;*/
>> + struct rte_eth_conf eth_conf;
>> + uint16_t port_id;
>> + int ret;
>> +
>> + mbuf_pool = rte_pktmbuf_pool_create("MBUF_POOL", NUM_MBUF,
>> MBUF_CACHE_SIZE, 0,
>> + RTE_MBUF_DEFAULT_BUF_SIZE, rte_socket_id());
>> +
>> + RTE_ETH_FOREACH_DEV(port_id) {
>> + memset(ð_conf, 0, sizeof(dev_info));
> This clears "eth_conf", but the sizeof is "dev_info". Why?
>
Will fix in next version, thx.
