> -----Original Message-----
> From: Wu, Jingjing <jingjing...@intel.com>
> Sent: Tuesday, March 19, 2024 10:15 PM
> To: dev@dpdk.org
> Cc: Wu, Jingjing <jingjing...@intel.com>; jer...@marvell.com;
> pbhagavat...@marvell.com; Yan, Zhirun <zhirun....@intel.com>;
> sta...@dpdk.org
> Subject: [PATCH] graph: fix head move when graph walk in mcore dispatch
> 
> Head move should happen after the core id check, otherwise source node will be
> missed.
> 
> Fixes: 35dfd9b9fd85 ("graph: introduce graph walk by cross-core dispatch")
> Cc: sta...@dpdk.org
> 
> Signed-off-by: Jingjing Wu <jingjing...@intel.com>
> ---
>  lib/graph/rte_graph_model_mcore_dispatch.h | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/lib/graph/rte_graph_model_mcore_dispatch.h
> b/lib/graph/rte_graph_model_mcore_dispatch.h
> index 75ec388cad..b96469296e 100644
> --- a/lib/graph/rte_graph_model_mcore_dispatch.h
> +++ b/lib/graph/rte_graph_model_mcore_dispatch.h
> @@ -97,12 +97,12 @@ rte_graph_walk_mcore_dispatch(struct rte_graph
> *graph)
>               __rte_graph_mcore_dispatch_sched_wq_process(graph);
> 
>       while (likely(head != graph->tail)) {
> -             node = (struct rte_node *)RTE_PTR_ADD(graph,
> cir_start[(int32_t)head++]);
> +             node = (struct rte_node *)RTE_PTR_ADD(graph,
> +cir_start[(int32_t)head]);
> 
>               /* skip the src nodes which not bind with current worker */
>               if ((int32_t)head < 0 && node->dispatch.lcore_id != graph-
> >dispatch.lcore_id)
>                       continue;
> -
> +             head++;
If current src node not bind with current core, It will go into infinite loop.
This line would have no chance to run.

>               /* Schedule the node until all task/objs are done */
>               if (node->dispatch.lcore_id != RTE_MAX_LCORE &&
>                   graph->dispatch.lcore_id != node->dispatch.lcore_id &&
> --
> 2.34.1

Reply via email to