From: Balazs Nemeth <bnemeth@redhat.com>
To: bnemeth@redhat.com, dev@dpdk.org
Cc: dsinghrawat@marvell.com, rmody@marvell.com,
jerinjacobk@gmail.com, irusskikh@marvell.com
Subject: [dpdk-dev] [PATCH v2 4/7] net/qede: free packets in bulk instead of one by one
Date: Fri, 26 Mar 2021 12:01:27 +0100 [thread overview]
Message-ID: <fd7ecedafded789bcf96f6f294ea79f08d73850e.1616754844.git.bnemeth@redhat.com> (raw)
In-Reply-To: <cover.1616754844.git.bnemeth@redhat.com>
rte_pktmbuf_free_bulk calls rte_mempool_put_bulk with the number of
pending packets to return to the mempool. In contrast, rte_pktmbuf_free
calls rte_mempool_put that calls rte_mempool_put_bulk with one object.
An important performance related downside of adding one packet at a time
to the mempool is that on each call, the per-core cache pointer needs to
be read from tls while a single rte_mempool_put_bulk only reads from the
tls once.
Signed-off-by: Balazs Nemeth <bnemeth@redhat.com>
Reviewed-by: Igor Russkikh <irusskikh@marvell.com>
---
drivers/net/qede/qede_rxtx.c | 12 +++++++++++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/drivers/net/qede/qede_rxtx.c b/drivers/net/qede/qede_rxtx.c
index 9294f79eb..f439ee056 100644
--- a/drivers/net/qede/qede_rxtx.c
+++ b/drivers/net/qede/qede_rxtx.c
@@ -893,6 +893,7 @@ qede_process_tx_compl(__rte_unused struct ecore_dev *edev,
struct rte_mbuf *mbuf;
uint16_t nb_segs;
uint16_t idx;
+ uint16_t first_idx;
rte_compiler_barrier();
sw_tx_cons = ecore_chain_get_cons_idx(&txq->tx_pbl);
@@ -907,6 +908,7 @@ qede_process_tx_compl(__rte_unused struct ecore_dev *edev,
remaining = hw_bd_cons - sw_tx_cons;
txq->nb_tx_avail += remaining;
+ first_idx = idx;
while (remaining) {
mbuf = txq->sw_tx_ring[idx];
@@ -921,11 +923,19 @@ qede_process_tx_compl(__rte_unused struct ecore_dev *edev,
nb_segs--;
}
- rte_pktmbuf_free(mbuf);
idx = (idx + 1) & mask;
PMD_TX_LOG(DEBUG, txq, "Freed tx packet\n");
}
txq->sw_tx_cons = idx;
+
+ if (first_idx > idx) {
+ rte_pktmbuf_free_bulk(&txq->sw_tx_ring[first_idx],
+ mask - first_idx + 1);
+ rte_pktmbuf_free_bulk(&txq->sw_tx_ring[0], idx);
+ } else {
+ rte_pktmbuf_free_bulk(&txq->sw_tx_ring[first_idx],
+ idx - first_idx);
+ }
}
static int qede_drain_txq(struct qede_dev *qdev,
--
2.30.2
next prev parent reply other threads:[~2021-03-26 11:02 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-26 11:01 [dpdk-dev] [PATCH v2 0/7] Optimize qede use of Rx/Tx entries Balazs Nemeth
2021-03-26 11:01 ` [dpdk-dev] [PATCH v2 1/7] net/qede: remove flags from Tx entry Balazs Nemeth
2021-03-26 11:01 ` [dpdk-dev] [PATCH v2 2/7] net/qede: get consumer index once Balazs Nemeth
2021-03-26 11:01 ` [dpdk-dev] [PATCH v2 3/7] net/qede: assume mbuf to free is never null Balazs Nemeth
2021-03-26 11:01 ` Balazs Nemeth [this message]
2021-03-26 11:01 ` [dpdk-dev] [PATCH v2 5/7] net/qede: prefetch hardware consumer Balazs Nemeth
2021-03-26 11:01 ` [dpdk-dev] [PATCH v2 6/7] net/qede: prefetch next packet to free Balazs Nemeth
2021-03-26 11:01 ` [dpdk-dev] [PATCH v2 7/7] net/qede: remove unnecessary field in Rx entry and simplify Balazs Nemeth
2021-03-27 14:02 ` [dpdk-dev] [PATCH v2 0/7] Optimize qede use of Rx/Tx entries Jerin Jacob
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=fd7ecedafded789bcf96f6f294ea79f08d73850e.1616754844.git.bnemeth@redhat.com \
--to=bnemeth@redhat.com \
--cc=dev@dpdk.org \
--cc=dsinghrawat@marvell.com \
--cc=irusskikh@marvell.com \
--cc=jerinjacobk@gmail.com \
--cc=rmody@marvell.com \
/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).