RE: [TEST] dpdk/app/test/test_mbuf.c test_refcnt_mbuf instability + fix proposal

[Julien Hascoet <jhascoet@kalrayinc.com>]

[-- Attachment #1.1: Type: text/plain, Size: 2267 bytes --]

Here is a patch, I let you amend it if needed as I'm a beginner in the dpdk project

Thanks
________________________________
De : David Marchand <david.marchand@redhat.com>
Envoyé : lundi 7 août 2023 09:26
À : Julien Hascoet <jhascoet@kalrayinc.com>; Olivier Matz <olivier.matz@6wind.com>
Cc : dev@dpdk.org <dev@dpdk.org>
Objet : Re: [TEST] dpdk/app/test/test_mbuf.c test_refcnt_mbuf instability + fix proposal

Hello Julien,

On Mon, Aug 7, 2023 at 8:19 AM Julien Hascoet <jhascoet@kalrayinc.com> wrote:
> from my understanding after debugging, in test_refcnt_iter the return value of rte_ring_enqueue is not checked; leading to lack of expected mbufs at the end checks.
>
> Here is some fix proposal that seems to work after running endurance tests for several days:
>
> diff --git a/app/test/test_mbuf.c b/app/test/test_mbuf.c
> index b4f436b5e2..8a5d26e4f6 100644
> --- a/app/test/test_mbuf.c
> +++ b/app/test/test_mbuf.c
> @@ -1033,12 +1033,17 @@ test_refcnt_iter(unsigned int lcore, unsigned int iter,
>                 tref += ref;
>                 if ((ref & 1) != 0) {
>                         rte_pktmbuf_refcnt_update(m, ref);
> -                       while (ref-- != 0)
> -                               rte_ring_enqueue(refcnt_mbuf_ring, m);
> +                       while (ref-- != 0) {
> +                               /* retry in case of failure */
> +                               while (rte_ring_enqueue(refcnt_mbuf_ring, m) != 0)
> +                                       ;
> +                       }
>                 } else {
>                         while (ref-- != 0) {
>                                 rte_pktmbuf_refcnt_update(m, 1);
> -                               rte_ring_enqueue(refcnt_mbuf_ring, m);
> +                               /* retry in case of failure */
> +                               while (rte_ring_enqueue(refcnt_mbuf_ring, m) != 0)
> +                                       ;
>                         }
>                 }
>                 rte_pktmbuf_free(m);
>
> Can you confirm ?

This analysis looks correct (though failing to enqueue in this unit
test seems strange to me).
Could you send a fix with a Fixes: line in the commitlog, and copying
the maintainer?

Thanks.


--
David Marchand








[-- Attachment #1.2: Type: text/html, Size: 6166 bytes --]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-app-fix-silent-enqueue-fail-in-test_mbuf-test_refcnt.patch --]
[-- Type: text/x-patch; name=0001-app-fix-silent-enqueue-fail-in-test_mbuf-test_refcnt.patch, Size: 1218 bytes --]

From 6b57da16f3eca121c91ac0c250c93d4d017eea55 Mon Sep 17 00:00:00 2001
From: jhascoet <jhascoet@kalray.eu>
Date: Mon, 7 Aug 2023 09:54:30 +0200
Subject: [PATCH] app: fix silent enqueue fail in test_mbuf test_refcnt_iter

In case of ring full state, we retry the enqueue
operation in order to avoid mbuf loss.

Fixes: af75078fece ("first public release")
---
 app/test/test_mbuf.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/app/test/test_mbuf.c b/app/test/test_mbuf.c
index efac01806b..be114e3302 100644
--- a/app/test/test_mbuf.c
+++ b/app/test/test_mbuf.c
@@ -1033,12 +1033,17 @@ test_refcnt_iter(unsigned int lcore, unsigned int iter,
 		tref += ref;
 		if ((ref & 1) != 0) {
 			rte_pktmbuf_refcnt_update(m, ref);
-			while (ref-- != 0)
-				rte_ring_enqueue(refcnt_mbuf_ring, m);
+			while (ref-- != 0) {
+				/* retry in case of failure */
+				while (rte_ring_enqueue(refcnt_mbuf_ring, m) != 0)
+					;
+			}
 		} else {
 			while (ref-- != 0) {
 				rte_pktmbuf_refcnt_update(m, 1);
-				rte_ring_enqueue(refcnt_mbuf_ring, m);
+				/* retry in case of failure */
+				while (rte_ring_enqueue(refcnt_mbuf_ring, m) != 0)
+					;
 			}
 		}
 		rte_pktmbuf_free(m);
-- 
2.34.1