From: "Morten Brørup" <mb@smartsharesystems.com>
To: dev@dpdk.org, Thomas Monjalon <thomas@monjalon.net>,
Stephen Hemminger <stephen@networkplumber.org>,
Bruce Richardson <bruce.richardson@intel.com>,
Konstantin Ananyev <konstantin.ananyev@huawei.com>,
Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>,
Ivan Malov <ivan.malov@arknetworks.am>,
Chengwen Feng <fengchengwen@huawei.com>
Cc: "Morten Brørup" <mb@smartsharesystems.com>
Subject: [PATCH v6 3/3] mbuf: no need to reset all fields on reinitialized mbufs
Date: Fri, 22 Aug 2025 12:47:31 +0000 [thread overview]
Message-ID: <20250822124731.20633-4-mb@smartsharesystems.com> (raw)
In-Reply-To: <20250822124731.20633-1-mb@smartsharesystems.com>
The 'next' and 'nb_segs' fields are already reset on newly allocated
reinitialized mbufs (a.k.a. raw mbufs), so a simpler reset function for
reinitialized mbufs was added.
The 'ol_flags' field must indicate whether the mbuf uses an external
buffer.
Unlike the normal packet mbuf reset function, which gets this information
from the mbuf itself (by masking the RTE_MBUF_F_EXTERNAL flag), the
simpler reset function gets this information from the mempool, thereby
reducing the number of read-modify-writes on the mbuf from two to one.
The remaining read-modify-write on the mbuf in the simpler reset function
is in rte_pktmbuf_reset_headroom(), where the 'buf_len' field is read
before writing the 'data_off' field.
And I am hoping that it will be possible to eliminate that one too.
Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
---
lib/mbuf/rte_mbuf.h | 43 ++++++++++++++++++++++++++++++++++++++-----
1 file changed, 38 insertions(+), 5 deletions(-)
diff --git a/lib/mbuf/rte_mbuf.h b/lib/mbuf/rte_mbuf.h
index 49c93ab356..92a1e8ba7e 100644
--- a/lib/mbuf/rte_mbuf.h
+++ b/lib/mbuf/rte_mbuf.h
@@ -954,6 +954,39 @@ static inline void rte_pktmbuf_reset_headroom(struct rte_mbuf *m)
(uint16_t)m->buf_len);
}
+/**
+ * Reset the fields of a packet mbuf to their default values.
+ *
+ * The caller must ensure that the mbuf comes from the specified mempool,
+ * is direct and properly reinitialized (refcnt=1, next=NULL, nb_segs=1),
+ * as done by rte_pktmbuf_prefree_seg().
+ *
+ * This function should be used with care, when optimization is required.
+ * For standard needs, prefer rte_pktmbuf_reset().
+ *
+ * @param mp
+ * The mempool to which the mbuf belongs.
+ * @param m
+ * The packet mbuf to be reset.
+ */
+static inline void rte_mbuf_raw_reset(struct rte_mempool *mp, struct rte_mbuf *m)
+{
+ uint32_t flags = rte_pktmbuf_priv_flags(mp);
+
+ m->pkt_len = 0;
+ m->tx_offload = 0;
+ m->vlan_tci = 0;
+ m->vlan_tci_outer = 0;
+ m->port = RTE_MBUF_PORT_INVALID;
+
+ m->ol_flags = (flags & RTE_PKTMBUF_POOL_F_PINNED_EXT_BUF) ? RTE_MBUF_F_EXTERNAL : 0;
+ m->packet_type = 0;
+ rte_pktmbuf_reset_headroom(m);
+
+ m->data_len = 0;
+ __rte_mbuf_sanity_check(m, 1);
+}
+
/**
* Reset the fields of a packet mbuf to their default values.
*
@@ -997,7 +1030,7 @@ static inline struct rte_mbuf *rte_pktmbuf_alloc(struct rte_mempool *mp)
{
struct rte_mbuf *m;
if ((m = rte_mbuf_raw_alloc(mp)) != NULL)
- rte_pktmbuf_reset(m);
+ rte_mbuf_raw_reset(mp, m);
return m;
}
@@ -1033,19 +1066,19 @@ static inline int rte_pktmbuf_alloc_bulk(struct rte_mempool *pool,
switch (count % 4) {
case 0:
while (idx != count) {
- rte_pktmbuf_reset(mbufs[idx]);
+ rte_mbuf_raw_reset(pool, mbufs[idx]);
idx++;
/* fall-through */
case 3:
- rte_pktmbuf_reset(mbufs[idx]);
+ rte_mbuf_raw_reset(pool, mbufs[idx]);
idx++;
/* fall-through */
case 2:
- rte_pktmbuf_reset(mbufs[idx]);
+ rte_mbuf_raw_reset(pool, mbufs[idx]);
idx++;
/* fall-through */
case 1:
- rte_pktmbuf_reset(mbufs[idx]);
+ rte_mbuf_raw_reset(pool, mbufs[idx]);
idx++;
/* fall-through */
}
--
2.43.0
prev parent reply other threads:[~2025-08-22 12:47 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-08-21 15:02 [PATCH v5 0/3] mbuf: simplify handling of " Morten Brørup
2025-08-21 15:02 ` [PATCH v5 1/3] mbuf: de-inline sanity checking a reinitialized mbuf Morten Brørup
2025-08-21 15:02 ` [PATCH v5 2/3] promote reinitialized mbuf free and alloc bulk functions as stable Morten Brørup
2025-08-21 15:02 ` [PATCH v5 3/3] mbuf: no need to reset all fields on reinitialized mbufs Morten Brørup
2025-08-22 12:47 ` [PATCH v6 0/3] mbuf: simplify handling of " Morten Brørup
2025-08-22 12:47 ` [PATCH v6 1/3] mbuf: de-inline sanity checking a reinitialized mbuf Morten Brørup
2025-08-22 14:26 ` Morten Brørup
2025-08-22 12:47 ` [PATCH v6 2/3] mbuf: promote raw free and alloc bulk functions as stable Morten Brørup
2025-08-22 12:47 ` Morten Brørup [this message]
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=20250822124731.20633-4-mb@smartsharesystems.com \
--to=mb@smartsharesystems.com \
--cc=andrew.rybchenko@oktetlabs.ru \
--cc=bruce.richardson@intel.com \
--cc=dev@dpdk.org \
--cc=fengchengwen@huawei.com \
--cc=ivan.malov@arknetworks.am \
--cc=konstantin.ananyev@huawei.com \
--cc=stephen@networkplumber.org \
--cc=thomas@monjalon.net \
/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).