All of lore.kernel.org
 help / color / mirror / Atom feed
From: Shailend Chand <shailend@google.com>
To: netdev@vger.kernel.org
Cc: almasrymina@google.com, davem@davemloft.net, edumazet@google.com,
	 hramamurthy@google.com, jeroendb@google.com, kuba@kernel.org,
	 pabeni@redhat.com, pkaligineedi@google.com, willemb@google.com,
	 Shailend Chand <shailend@google.com>
Subject: [PATCH net-next 05/10] gve: Make gve_turnup work for nonempty queues
Date: Tue, 30 Apr 2024 23:14:14 +0000	[thread overview]
Message-ID: <20240430231420.699177-6-shailend@google.com> (raw)
In-Reply-To: <20240430231420.699177-1-shailend@google.com>

gVNIC has a requirement that all queues have to be quiesced before any
queue is operated on (created or destroyed). To enable the
implementation of future ndo hooks that work on a single queue, we need
to evolve gve_turnup to account for queues already having some
unprocessed descriptors in the ring.

Say rxq 4 is being stopped and started via the queue api. Due to gve's
requirement of quiescence, queues 0 through 3 are not processing their
rings while queue 4 is being toggled. Once they are made live, these
queues need to be poked to cause them to check their rings for
descriptors that were written during their brief period of quiescence.

Tested-by: Mina Almasry <almasrymina@google.com>
Reviewed-by: Praveen Kaligineedi <pkaligineedi@google.com>
Reviewed-by: Harshitha Ramamurthy <hramamurthy@google.com>
Signed-off-by: Shailend Chand <shailend@google.com>
---
 drivers/net/ethernet/google/gve/gve_main.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c
index 469a914c71d6..ef902b72b9a9 100644
--- a/drivers/net/ethernet/google/gve/gve_main.c
+++ b/drivers/net/ethernet/google/gve/gve_main.c
@@ -1979,6 +1979,13 @@ static void gve_turnup(struct gve_priv *priv)
 			gve_set_itr_coalesce_usecs_dqo(priv, block,
 						       priv->tx_coalesce_usecs);
 		}
+
+		/* Any descs written by the NIC before this barrier will be
+		 * handled by the one-off napi schedule below. Whereas any
+		 * descs after the barrier will generate interrupts.
+		 */
+		mb();
+		napi_schedule(&block->napi);
 	}
 	for (idx = 0; idx < priv->rx_cfg.num_queues; idx++) {
 		int ntfy_idx = gve_rx_idx_to_ntfy(priv, idx);
@@ -1994,6 +2001,13 @@ static void gve_turnup(struct gve_priv *priv)
 			gve_set_itr_coalesce_usecs_dqo(priv, block,
 						       priv->rx_coalesce_usecs);
 		}
+
+		/* Any descs written by the NIC before this barrier will be
+		 * handled by the one-off napi schedule below. Whereas any
+		 * descs after the barrier will generate interrupts.
+		 */
+		mb();
+		napi_schedule(&block->napi);
 	}
 
 	gve_set_napi_enabled(priv);
-- 
2.45.0.rc0.197.gbae5840b3b-goog


  parent reply	other threads:[~2024-04-30 23:14 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-04-30 23:14 [PATCH net-next 00/10] gve: Implement queue api Shailend Chand
2024-04-30 23:14 ` [PATCH net-next 01/10] queue_api: define " Shailend Chand
2024-04-30 23:14 ` [PATCH net-next 02/10] gve: Make the GQ RX free queue funcs idempotent Shailend Chand
2024-04-30 23:14 ` [PATCH net-next 03/10] gve: Add adminq funcs to add/remove a single Rx queue Shailend Chand
2024-05-01  4:19   ` Ratheesh Kannoth
2024-05-01 13:50   ` Willem de Bruijn
2024-05-01 23:27     ` Shailend Chand
2024-05-02  1:39       ` Willem de Bruijn
2024-04-30 23:14 ` [PATCH net-next 04/10] gve: Make gve_turn(up|down) ignore stopped queues Shailend Chand
2024-04-30 23:14 ` Shailend Chand [this message]
2024-04-30 23:14 ` [PATCH net-next 06/10] gve: Avoid rescheduling napi if on wrong cpu Shailend Chand
2024-04-30 23:14 ` [PATCH net-next 07/10] gve: Reset Rx ring state in the ring-stop funcs Shailend Chand
2024-04-30 23:14 ` [PATCH net-next 08/10] gve: Account for stopped queues when reading NIC stats Shailend Chand
2024-04-30 23:14 ` [PATCH net-next 09/10] gve: Alloc and free QPLs with the rings Shailend Chand
2024-05-01 15:31   ` Jakub Kicinski
2024-04-30 23:14 ` [PATCH net-next 10/10] gve: Implement queue api Shailend Chand

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=20240430231420.699177-6-shailend@google.com \
    --to=shailend@google.com \
    --cc=almasrymina@google.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=hramamurthy@google.com \
    --cc=jeroendb@google.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=pkaligineedi@google.com \
    --cc=willemb@google.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.