[PATCH] app/dma_perf: skip case if worker maps to main lcore

[PATCH] app/dma_perf: skip case if worker maps to main lcore

[Rupesh Chiluka]

Refuse to run DMA/CPU mem-copy cases when any worker is bound to the
EAL main lcore.

Signed-off-by: Rupesh Chiluka <rchiluka@marvell.com>
---
 app/test-dma-perf/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/app/test-dma-perf/main.c b/app/test-dma-perf/main.c
index 4249dcfd3d..b6aa5b8401 100644
--- a/app/test-dma-perf/main.c
+++ b/app/test-dma-perf/main.c
@@ -109,6 +109,7 @@ run_test_case(struct test_configure *case_cfg)
 static void
 run_test(uint32_t case_id, struct test_configure *case_cfg)
 {
+	uint32_t main_lcore = rte_get_main_lcore();
 	uint32_t nb_lcores = rte_lcore_count();
 	struct test_configure_entry *mem_size = &case_cfg->mem_size;
 	struct test_configure_entry *buf_size = &case_cfg->buf_size;
@@ -122,6 +123,14 @@ run_test(uint32_t case_id, struct test_configure *case_cfg)
 		return;
 	}
 
+	for (uint32_t i = 0; i < case_cfg->num_worker; i++) {
+		if (case_cfg->dma_config[i].lcore_dma_map.lcore == main_lcore) {
+			printf("Case %u: worker %u cannot run on the EAL main lcore (%u).\n",
+			       case_id, i, main_lcore);
+			return;
+		}
+	}
+
 	printf("Number of used lcores: %u.\n", nb_lcores);
 
 	if (mem_size->incr != 0)
-- 
2.48.1

Re: [PATCH] app/dma_perf: skip case if worker maps to main lcore

[Bruce Richardson]

On Tue, Jun 23, 2026 at 10:28:41AM +0530, Rupesh Chiluka wrote:
> Refuse to run DMA/CPU mem-copy cases when any worker is bound to the
> EAL main lcore.
> 

Can you explain a bit more why?

> Signed-off-by: Rupesh Chiluka <rchiluka@marvell.com>
> ---
>  app/test-dma-perf/main.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/app/test-dma-perf/main.c b/app/test-dma-perf/main.c
> index 4249dcfd3d..b6aa5b8401 100644
> --- a/app/test-dma-perf/main.c
> +++ b/app/test-dma-perf/main.c
> @@ -109,6 +109,7 @@ run_test_case(struct test_configure *case_cfg)
>  static void
>  run_test(uint32_t case_id, struct test_configure *case_cfg)
>  {
> +	uint32_t main_lcore = rte_get_main_lcore();
>  	uint32_t nb_lcores = rte_lcore_count();
>  	struct test_configure_entry *mem_size = &case_cfg->mem_size;
>  	struct test_configure_entry *buf_size = &case_cfg->buf_size;
> @@ -122,6 +123,14 @@ run_test(uint32_t case_id, struct test_configure *case_cfg)
>  		return;
>  	}
>  
> +	for (uint32_t i = 0; i < case_cfg->num_worker; i++) {
> +		if (case_cfg->dma_config[i].lcore_dma_map.lcore == main_lcore) {
> +			printf("Case %u: worker %u cannot run on the EAL main lcore (%u).\n",
> +			       case_id, i, main_lcore);
> +			return;
> +		}
> +	}
> +
>  	printf("Number of used lcores: %u.\n", nb_lcores);
>  
>  	if (mem_size->incr != 0)
> -- 
> 2.48.1
> 

Re: [PATCH] app/dma_perf: skip case if worker maps to main lcore

[fengchengwen]

On 6/23/2026 4:07 PM, Bruce Richardson wrote:
> On Tue, Jun 23, 2026 at 10:28:41AM +0530, Rupesh Chiluka wrote:
>> Refuse to run DMA/CPU mem-copy cases when any worker is bound to the
>> EAL main lcore.
>>
> 
> Can you explain a bit more why?

+1
I understand this bugfix, as the following config, the test will stuck
because the master lcore is the same with worker lcore.

[GLOBAL]
eal_args=--in-memory --file-prefix=test -l 10-12

[case4]
type=CPU_MEM_COPY
mem_size=10
buf_size=64,8192,2,MUL
src_numa_node=0
dst_numa_node=1
lcore = 10, 11

So please add more explain as Bruce suggest, and also add Fixes/CC tag

> 
>> Signed-off-by: Rupesh Chiluka <rchiluka@marvell.com>
>> ---
>>  app/test-dma-perf/main.c | 9 +++++++++
>>  1 file changed, 9 insertions(+)
>>
>> diff --git a/app/test-dma-perf/main.c b/app/test-dma-perf/main.c
>> index 4249dcfd3d..b6aa5b8401 100644
>> --- a/app/test-dma-perf/main.c
>> +++ b/app/test-dma-perf/main.c
>> @@ -109,6 +109,7 @@ run_test_case(struct test_configure *case_cfg)
>>  static void
>>  run_test(uint32_t case_id, struct test_configure *case_cfg)
>>  {
>> +	uint32_t main_lcore = rte_get_main_lcore();
>>  	uint32_t nb_lcores = rte_lcore_count();
>>  	struct test_configure_entry *mem_size = &case_cfg->mem_size;
>>  	struct test_configure_entry *buf_size = &case_cfg->buf_size;
>> @@ -122,6 +123,14 @@ run_test(uint32_t case_id, struct test_configure *case_cfg)
>>  		return;
>>  	}
>>  
>> +	for (uint32_t i = 0; i < case_cfg->num_worker; i++) {

Please define the variables in front of function impl.

Thanks

>> +		if (case_cfg->dma_config[i].lcore_dma_map.lcore == main_lcore) {
>> +			printf("Case %u: worker %u cannot run on the EAL main lcore (%u).\n",
>> +			       case_id, i, main_lcore);
>> +			return;
>> +		}
>> +	}
>> +
>>  	printf("Number of used lcores: %u.\n", nb_lcores);
>>  
>>  	if (mem_size->incr != 0)
>> -- 
>> 2.48.1
>>
> 

Re: [PATCH] app/dma_perf: skip case if worker maps to main lcore

[Stephen Hemminger]

On Wed, 24 Jun 2026 08:57:56 +0800
fengchengwen <fengchengwen@huawei.com> wrote:

> >>  
> >> +	for (uint32_t i = 0; i < case_cfg->num_worker; i++) {  
> 
> Please define the variables in front of function impl.

No longer required in DPDK coding style.

[PATCH] app/dma_perf: skip case if worker maps to main lcore

[Rupesh Chiluka]

When a configured worker lcore is same as EAL main lcore, the main
thread calls `rte_eal_remote_launch()` on itself and then waits for
`worker_info.ready_flag`. The worker function cannot run on the main
lcore while the main thread is blocked in that loop, and the test hangs.

Reject such cases early in the `run_test()`.

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

Signed-off-by: Rupesh Chiluka <rchiluka@marvell.com>
---
 app/test-dma-perf/main.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/app/test-dma-perf/main.c b/app/test-dma-perf/main.c
index 4249dcfd3d..a22434c558 100644
--- a/app/test-dma-perf/main.c
+++ b/app/test-dma-perf/main.c
@@ -109,7 +109,9 @@ run_test_case(struct test_configure *case_cfg)
 static void
 run_test(uint32_t case_id, struct test_configure *case_cfg)
 {
+	uint32_t worker_lcore;
 	uint32_t nb_lcores = rte_lcore_count();
+	uint32_t main_lcore = rte_get_main_lcore();
 	struct test_configure_entry *mem_size = &case_cfg->mem_size;
 	struct test_configure_entry *buf_size = &case_cfg->buf_size;
 	struct test_configure_entry *ring_size = &case_cfg->ring_size;
@@ -122,6 +124,15 @@ run_test(uint32_t case_id, struct test_configure *case_cfg)
 		return;
 	}
 
+	for (uint32_t i = 0; i < case_cfg->num_worker; i++) {
+		worker_lcore = case_cfg->dma_config[i].lcore_dma_map.lcore;
+		if (worker_lcore == main_lcore) {
+			printf("Case %u: worker %u (lcore %u) cannot run on the EAL main lcore (%u).\n",
+			       case_id, i, worker_lcore, main_lcore);
+			return;
+		}
+	}
+
 	printf("Number of used lcores: %u.\n", nb_lcores);
 
 	if (mem_size->incr != 0)
-- 
2.48.1

[PATCH v2] app/dma_perf: skip case if worker maps to main lcore

[Rupesh Chiluka]

When a configured worker lcore is same as EAL main lcore, the main
thread calls `rte_eal_remote_launch()` on itself and then waits for
`worker_info.ready_flag`. The worker function cannot run on the main
lcore while the main thread is blocked in that loop, and the test hangs.

Reject such cases early in the `run_test()`.

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

Signed-off-by: Rupesh Chiluka <rchiluka@marvell.com>
---
 app/test-dma-perf/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/app/test-dma-perf/main.c b/app/test-dma-perf/main.c
index 4249dcfd3d..2809e5417c 100644
--- a/app/test-dma-perf/main.c
+++ b/app/test-dma-perf/main.c
@@ -110,6 +110,7 @@ static void
 run_test(uint32_t case_id, struct test_configure *case_cfg)
 {
 	uint32_t nb_lcores = rte_lcore_count();
+	uint32_t main_lcore = rte_get_main_lcore();
 	struct test_configure_entry *mem_size = &case_cfg->mem_size;
 	struct test_configure_entry *buf_size = &case_cfg->buf_size;
 	struct test_configure_entry *ring_size = &case_cfg->ring_size;
@@ -122,6 +123,14 @@ run_test(uint32_t case_id, struct test_configure *case_cfg)
 		return;
 	}
 
+	for (uint32_t i = 0; i < case_cfg->num_worker; i++) {
+		if (case_cfg->dma_config[i].lcore_dma_map.lcore == main_lcore) {
+			printf("Case %u: worker %u cannot run on the EAL main lcore (%u).\n",
+			       case_id, i, main_lcore);
+			return;
+		}
+	}
+
 	printf("Number of used lcores: %u.\n", nb_lcores);
 
 	if (mem_size->incr != 0)
-- 
2.48.1

Re: [PATCH v2] app/dma_perf: skip case if worker maps to main lcore

[fengchengwen]

Acked-by: Chengwen Feng <fengchengwen@huawei.com>

On 6/24/2026 1:42 PM, Rupesh Chiluka wrote:
> When a configured worker lcore is same as EAL main lcore, the main
> thread calls `rte_eal_remote_launch()` on itself and then waits for
> `worker_info.ready_flag`. The worker function cannot run on the main
> lcore while the main thread is blocked in that loop, and the test hangs.
> 
> Reject such cases early in the `run_test()`.
> 
> Fixes: 623dc9364dc6 ("app/dma-perf: introduce DMA performance test")
> CC: stable@dpdk.org
> 
> Signed-off-by: Rupesh Chiluka <rchiluka@marvell.com>