[EXTERNAL] [PATCH] graph: optimize graph search when scheduling nodes

[Jerin Jacob]

> -----Original Message-----
> From: Huichao Cai <chcchc88 at 163.com>
> Sent: Friday, November 8, 2024 7:10 AM
> To: Jerin Jacob <jerinj at marvell.com>
> Cc: Kiran Kumar Kokkilagadda <kirankumark at marvell.com>; Nithin Kumar
> Dabilpuram <ndabilpuram at marvell.com>; yanzhirun_163 at 163.com;
> dev at dpdk.org
> Subject: Re:RE: [EXTERNAL] [PATCH] graph: optimize graph search when
> scheduling nodes
> 
> > Is n't breaking the ABI? So can't we modify the ABI, or is there any
> > special operation required to modify the ABI? ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍
> > ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍
> > ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍
> > ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍ ‍
> 
> ZjQcmQRYFpfptBannerEnd
> 
> > Is n't breaking the ABI?
> 
> So can't we modify the ABI, or is there any special operation required to modify
> the ABI?

Only LTS release (xx.11) can change the ABI after sending deprecation notice.
Looking at the pahole output, one option will be making dispatch and new semi fastpath
Additions like  xstat_off can be min cache aligned to make room for future expansion and 
to make sure have better performance.

For xstat_off addition, there was deprecation notice to update rte_node.
If there are no objection, may be we can try following in this release to not wait
Huichao for one more year.


[main] [dpdk.org] $ git diff
diff --git a/lib/graph/rte_graph_worker_common.h b/lib/graph/rte_graph_worker_common.h
index a518af2b2a..ec9a82186d 100644
--- a/lib/graph/rte_graph_worker_common.h
+++ b/lib/graph/rte_graph_worker_common.h
@@ -104,6 +104,7 @@ struct __rte_cache_aligned rte_node {
        /** Original process function when pcap is enabled. */
        rte_node_process_t original_process;

+       alignas(RTE_CACHE_LINE_MIN_SIZE)
        union {
                /* Fast schedule area for mcore dispatch model */
                struct {
@@ -112,6 +113,7 @@ struct __rte_cache_aligned rte_node {
                        uint64_t total_sched_fail; /**< Number of scheduled failure. */
                } dispatch;
        };
+       alignas(RTE_CACHE_LINE_MIN_SIZE)
        rte_graph_off_t xstat_off; /**< Offset to xstat counters. */
        /* Fast path area  */
        __extension__ struct __rte_cache_aligned {