* [PATCH 20.11] net/nfp: free HW rings memzone on queue release
@ 2022-02-22 13:37 heinrich.kuhn
2022-02-22 14:27 ` Luca Boccassi
0 siblings, 1 reply; 2+ messages in thread
From: heinrich.kuhn @ 2022-02-22 13:37 UTC (permalink / raw)
To: stable; +Cc: Heinrich Kuhn, Simon Horman
From: Heinrich Kuhn <heinrich.kuhn@corigine.com>
[upstream commit 15174c40b70dfa89c1d94ff93ae756fae8d471bc]
During rx/tx queue setup, memory is reserved for the hardware rings.
This memory zone should subsequently be freed in the queue release
logic. This commit also adds a call to the release logic in the
dev_close() callback so that the ring memzone may be freed during port
close too.
Fixes: b812daadad0d ("nfp: add Rx and Tx")
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
drivers/net/nfp/nfp_net.c | 16 ++++++++++++++++
drivers/net/nfp/nfp_net_pmd.h | 6 ++++++
2 files changed, 22 insertions(+)
diff --git a/drivers/net/nfp/nfp_net.c b/drivers/net/nfp/nfp_net.c
index df3b7711dc..fa42548e7a 100644
--- a/drivers/net/nfp/nfp_net.c
+++ b/drivers/net/nfp/nfp_net.c
@@ -223,6 +223,7 @@ nfp_net_rx_queue_release(void *rx_queue)
if (rxq) {
nfp_net_rx_queue_release_mbufs(rxq);
+ rte_memzone_free(rxq->tz);
rte_free(rxq->rxbufs);
rte_free(rxq);
}
@@ -259,6 +260,7 @@ nfp_net_tx_queue_release(void *tx_queue)
if (txq) {
nfp_net_tx_queue_release_mbufs(txq);
+ rte_memzone_free(txq->tz);
rte_free(txq->txbufs);
rte_free(txq);
}
@@ -888,11 +890,15 @@ nfp_net_close(struct rte_eth_dev *dev)
for (i = 0; i < dev->data->nb_tx_queues; i++) {
nfp_net_reset_tx_queue(
(struct nfp_net_txq *)dev->data->tx_queues[i]);
+ nfp_net_tx_queue_release(
+ (struct nfp_net_txq *)dev->data->tx_queues[i]);
}
for (i = 0; i < dev->data->nb_rx_queues; i++) {
nfp_net_reset_rx_queue(
(struct nfp_net_rxq *)dev->data->rx_queues[i]);
+ nfp_net_rx_queue_release(
+ (struct nfp_net_rxq *)dev->data->rx_queues[i]);
}
rte_intr_disable(&pci_dev->intr_handle);
@@ -1609,6 +1615,11 @@ nfp_net_rx_queue_setup(struct rte_eth_dev *dev,
rxq->dma = (uint64_t)tz->iova;
rxq->rxds = (struct nfp_net_rx_desc *)tz->addr;
+ /* Also save the pointer to the memzone struct so it can be freed
+ * if needed
+ */
+ rxq->tz = tz;
+
/* mbuf pointers array for referencing mbufs linked to RX descriptors */
rxq->rxbufs = rte_zmalloc_socket("rxq->rxbufs",
sizeof(*rxq->rxbufs) * nb_desc,
@@ -1749,6 +1760,11 @@ nfp_net_tx_queue_setup(struct rte_eth_dev *dev, uint16_t queue_idx,
return -ENOMEM;
}
+ /* Save the pointer to the memzone struct so it can be freed
+ * if needed
+ */
+ txq->tz = tz;
+
txq->tx_count = nb_desc;
txq->tx_free_thresh = tx_free_thresh;
txq->tx_pthresh = tx_conf->tx_thresh.pthresh;
diff --git a/drivers/net/nfp/nfp_net_pmd.h b/drivers/net/nfp/nfp_net_pmd.h
index ac6118d639..67393f611e 100644
--- a/drivers/net/nfp/nfp_net_pmd.h
+++ b/drivers/net/nfp/nfp_net_pmd.h
@@ -235,6 +235,9 @@ struct nfp_net_txq {
*/
struct nfp_net_tx_desc *txds;
+ /* Pointer to the memzone for the ring */
+ const struct rte_memzone *tz;
+
/*
* At this point 48 bytes have been used for all the fields in the
* TX critical path. We have room for 8 bytes and still all placed
@@ -370,6 +373,9 @@ struct nfp_net_rxq {
/* DMA address of the queue */
__le64 dma;
+ /* Pointer to the memzone for the ring */
+ const struct rte_memzone *tz;
+
/*
* Queue information: @qidx is the queue index from Linux's
* perspective. @fl_qcidx is the index of the Queue
--
2.27.0
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH 20.11] net/nfp: free HW rings memzone on queue release
2022-02-22 13:37 [PATCH 20.11] net/nfp: free HW rings memzone on queue release heinrich.kuhn
@ 2022-02-22 14:27 ` Luca Boccassi
0 siblings, 0 replies; 2+ messages in thread
From: Luca Boccassi @ 2022-02-22 14:27 UTC (permalink / raw)
To: heinrich.kuhn, stable; +Cc: Simon Horman
On Tue, 2022-02-22 at 15:37 +0200, heinrich.kuhn@corigine.com wrote:
> From: Heinrich Kuhn <heinrich.kuhn@corigine.com>
>
> [upstream commit 15174c40b70dfa89c1d94ff93ae756fae8d471bc]
>
> During rx/tx queue setup, memory is reserved for the hardware rings.
> This memory zone should subsequently be freed in the queue release
> logic. This commit also adds a call to the release logic in the
> dev_close() callback so that the ring memzone may be freed during port
> close too.
>
> Fixes: b812daadad0d ("nfp: add Rx and Tx")
>
> Signed-off-by: Heinrich Kuhn <heinrich.kuhn@corigine.com>
> Signed-off-by: Simon Horman <simon.horman@corigine.com>
> ---
> drivers/net/nfp/nfp_net.c | 16 ++++++++++++++++
> drivers/net/nfp/nfp_net_pmd.h | 6 ++++++
> 2 files changed, 22 insertions(+)
Thanks, applied and pushed.
--
Kind regards,
Luca Boccassi
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-02-22 14:27 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-02-22 13:37 [PATCH 20.11] net/nfp: free HW rings memzone on queue release heinrich.kuhn
2022-02-22 14:27 ` Luca Boccassi
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).