DPDK patches and discussions
 help / color / mirror / Atom feed
From: Chengwen Feng <fengchengwen@huawei.com>
To: <thomas@monjalon.net>, <honest.jiang@foxmail.com>
Cc: <dev@dpdk.org>, <liuyonglong@huawei.com>, <vattunuru@marvell.com>
Subject: [PATCH v3 12/13] app/dma-perf: fix on-flight DMA when verify data
Date: Mon, 13 Oct 2025 11:02:35 +0800	[thread overview]
Message-ID: <20251013030236.3861-13-fengchengwen@huawei.com> (raw)
In-Reply-To: <20251013030236.3861-1-fengchengwen@huawei.com>

There maybe on-flight DMA when verify_data() because the DMA device
may still working when worker exit.

This commit add wait DMA complete stage before worker exit.

Fixes: 623dc9364dc6 ("app/dma-perf: introduce DMA performance test")
Cc: stable@dpdk.org

Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
---
 app/test-dma-perf/benchmark.c | 54 +++++++++++++++++++++++------------
 1 file changed, 35 insertions(+), 19 deletions(-)

diff --git a/app/test-dma-perf/benchmark.c b/app/test-dma-perf/benchmark.c
index ab2bf0cbf5..8e755c2afa 100644
--- a/app/test-dma-perf/benchmark.c
+++ b/app/test-dma-perf/benchmark.c
@@ -19,7 +19,6 @@
 #define MAX_DMA_CPL_NB 255
 
 #define TEST_WAIT_U_SECOND 10000
-#define POLL_MAX 1000
 
 #define CSV_LINE_DMA_FMT "Scenario %u,%u,%s,%u,%u,%u,%u,%.2lf,%" PRIu64 ",%.3lf,%.3lf\n"
 #define CSV_LINE_CPU_FMT "Scenario %u,%u,NA,NA,NA,%u,%u,%.2lf,%" PRIu64 ",%.3lf,%.3lf\n"
@@ -282,6 +281,39 @@ do_dma_submit_and_poll(uint16_t dev_id, uint64_t *async_cnt,
 	worker_info->total_cpl += nr_cpl;
 }
 
+static int
+do_dma_submit_and_wait_cpl(uint16_t dev_id, uint64_t async_cnt)
+{
+#define MAX_WAIT_MSEC	1000
+#define MAX_POLL	1000
+	enum rte_dma_vchan_status st;
+	uint32_t poll_cnt = 0;
+	uint32_t wait_ms = 0;
+	uint16_t nr_cpl;
+
+	rte_dma_submit(dev_id, 0);
+
+	if (rte_dma_vchan_status(dev_id, 0, &st) < 0) {
+		rte_delay_ms(MAX_WAIT_MSEC);
+		goto wait_cpl;
+	}
+
+	while (st == RTE_DMA_VCHAN_ACTIVE && wait_ms++ < MAX_WAIT_MSEC) {
+		rte_delay_ms(1);
+		rte_dma_vchan_status(dev_id, 0, &st);
+	}
+
+wait_cpl:
+	while ((async_cnt > 0) && (poll_cnt++ < MAX_POLL)) {
+		nr_cpl = rte_dma_completed(dev_id, 0, MAX_DMA_CPL_NB, NULL, NULL);
+		async_cnt -= nr_cpl;
+	}
+	if (async_cnt > 0)
+		PRINT_ERR("Error: wait DMA %u failed!\n", dev_id);
+
+	return async_cnt == 0 ? 0 : -1;
+}
+
 static inline int
 do_dma_plain_mem_copy(void *p)
 {
@@ -293,10 +325,8 @@ do_dma_plain_mem_copy(void *p)
 	const uint32_t buf_size = para->buf_size;
 	struct rte_mbuf **srcs = para->srcs;
 	struct rte_mbuf **dsts = para->dsts;
-	uint16_t nr_cpl;
 	uint64_t async_cnt = 0;
 	uint32_t i;
-	uint32_t poll_cnt = 0;
 	int ret;
 
 	worker_info->stop_flag = false;
@@ -327,13 +357,7 @@ do_dma_plain_mem_copy(void *p)
 			break;
 	}
 
-	rte_dma_submit(dev_id, 0);
-	while ((async_cnt > 0) && (poll_cnt++ < POLL_MAX)) {
-		nr_cpl = rte_dma_completed(dev_id, 0, MAX_DMA_CPL_NB, NULL, NULL);
-		async_cnt -= nr_cpl;
-	}
-
-	return 0;
+	return do_dma_submit_and_wait_cpl(dev_id, async_cnt);
 }
 
 static inline int
@@ -349,8 +373,6 @@ do_dma_sg_mem_copy(void *p)
 	const uint16_t dev_id = para->dev_id;
 	uint32_t nr_buf = para->nr_buf;
 	uint64_t async_cnt = 0;
-	uint32_t poll_cnt = 0;
-	uint16_t nr_cpl;
 	uint32_t i, j;
 	int ret;
 
@@ -386,13 +408,7 @@ do_dma_sg_mem_copy(void *p)
 			break;
 	}
 
-	rte_dma_submit(dev_id, 0);
-	while ((async_cnt > 0) && (poll_cnt++ < POLL_MAX)) {
-		nr_cpl = rte_dma_completed(dev_id, 0, MAX_DMA_CPL_NB, NULL, NULL);
-		async_cnt -= nr_cpl;
-	}
-
-	return 0;
+	return do_dma_submit_and_wait_cpl(dev_id, async_cnt);
 }
 
 static inline int
-- 
2.17.1


  parent reply	other threads:[~2025-10-13  3:04 UTC|newest]

Thread overview: 49+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-08-11 10:54 [PATCH 0/9] bugfix and refactor of dma-perf Chengwen Feng
2025-08-11 10:54 ` [PATCH 1/9] app/dma-perf: fix use-after-free Chengwen Feng
2025-08-11 10:54 ` [PATCH 2/9] app/dma-perf: add global section for config file Chengwen Feng
2025-08-11 10:54 ` [PATCH 3/9] app/dma-perf: use argparse lib to parse argument Chengwen Feng
2025-08-11 10:54 ` [PATCH 4/9] app/dma-perf: refactor output csv Chengwen Feng
2025-08-11 10:54 ` [PATCH 5/9] app/dma-perf: support list DMA devices Chengwen Feng
2025-08-11 10:54 ` [PATCH 6/9] app/dma-perf: add more global config Chengwen Feng
2025-08-11 10:54 ` [PATCH 7/9] app/dma-perf: remove invalid or redundant field Chengwen Feng
2025-08-11 10:54 ` [PATCH 8/9] app/dma-perf: refactor load config function Chengwen Feng
2025-08-11 10:54 ` [PATCH 9/9] app/dma-perf: refactor benchmark function Chengwen Feng
2025-08-12  2:06 ` [PATCH 00/10] bugfix and refactor of dma-perf Chengwen Feng
2025-08-12  2:06   ` [PATCH 01/10] app/dma-perf: fix use-after-free Chengwen Feng
2025-08-12  2:07   ` [PATCH 02/10] app/dma-perf: add global section for config file Chengwen Feng
2025-08-12  2:07   ` [PATCH 03/10] app/dma-perf: use argparse lib to parse argument Chengwen Feng
2025-08-12  2:07   ` [PATCH 04/10] app/dma-perf: refactor output csv Chengwen Feng
2025-08-12  2:07   ` [PATCH 05/10] app/dma-perf: support list DMA devices Chengwen Feng
2025-08-12  2:07   ` [PATCH 06/10] app/dma-perf: add more global config Chengwen Feng
2025-08-12  2:07   ` [PATCH 07/10] app/dma-perf: remove invalid or redundant field Chengwen Feng
2025-08-12  2:07   ` [PATCH 08/10] app/dma-perf: refactor load config function Chengwen Feng
2025-08-12  2:07   ` [PATCH 09/10] app/dma-perf: refactor benchmark function Chengwen Feng
2025-08-12  2:07   ` [PATCH 10/10] app/dma-perf: support specific error info Chengwen Feng
2025-09-11 14:53   ` [EXTERNAL] [PATCH 00/10] bugfix and refactor of dma-perf Vamsi Krishna Attunuru
2025-09-17  3:33 ` [PATCH v2 " Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 01/10] app/dma-perf: fix use-after-free Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 02/10] app/dma-perf: add global section for config file Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 03/10] app/dma-perf: use argparse lib to parse argument Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 04/10] app/dma-perf: refactor output csv Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 05/10] app/dma-perf: support list DMA devices Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 06/10] app/dma-perf: add more global config Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 07/10] app/dma-perf: remove invalid or redundant field Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 08/10] app/dma-perf: refactor load config function Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 09/10] app/dma-perf: refactor benchmark function Chengwen Feng
2025-09-17  3:33   ` [PATCH v2 10/10] app/dma-perf: support specific error info Chengwen Feng
2025-10-13  3:02 ` [PATCH v3 00/13] bugfix and refactor of dma-perf Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 01/13] app/dma-perf: fix use-after-free Chengwen Feng
2025-10-13  7:55     ` Bruce Richardson
2025-10-13  8:15       ` fengchengwen
2025-10-13  3:02   ` [PATCH v3 02/13] app/dma-perf: add global section for config file Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 03/13] app/dma-perf: use argparse lib to parse argument Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 04/13] app/dma-perf: refactor output csv Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 05/13] app/dma-perf: support list DMA devices Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 06/13] app/dma-perf: add more global config Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 07/13] app/dma-perf: remove invalid or redundant field Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 08/13] app/dma-perf: refactor load config function Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 09/13] app/dma-perf: refactor benchmark function Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 10/13] app/dma-perf: support specific error info Chengwen Feng
2025-10-13  3:02   ` [PATCH v3 11/13] app/dma-perf: fix segment fault with large size Chengwen Feng
2025-10-13  3:02   ` Chengwen Feng [this message]
2025-10-13  3:02   ` [PATCH v3 13/13] app/dma-perf: fix wrong stage to stop dmadev Chengwen Feng

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20251013030236.3861-13-fengchengwen@huawei.com \
    --to=fengchengwen@huawei.com \
    --cc=dev@dpdk.org \
    --cc=honest.jiang@foxmail.com \
    --cc=liuyonglong@huawei.com \
    --cc=thomas@monjalon.net \
    --cc=vattunuru@marvell.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).