linux-rdma.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Long Li <longli@microsoft.com>
To: Long Li <longli@microsoft.com>,
	Konstantin Taranov <kotaranov@microsoft.com>,
	Jakub Kicinski <kuba@kernel.org>,
	"David S . Miller" <davem@davemloft.net>,
	Paolo Abeni <pabeni@redhat.com>,
	Eric Dumazet <edumazet@google.com>,
	Andrew Lunn <andrew+netdev@lunn.ch>,
	Jason Gunthorpe <jgg@ziepe.ca>, Leon Romanovsky <leon@kernel.org>,
	Haiyang Zhang <haiyangz@microsoft.com>,
	"K . Y . Srinivasan" <kys@microsoft.com>,
	Wei Liu <wei.liu@kernel.org>, Dexuan Cui <decui@microsoft.com>,
	shradhagupta@linux.microsoft.com, Simon Horman <horms@kernel.org>,
	ernis@linux.microsoft.com, stephen@networkplumber.org
Cc: netdev@vger.kernel.org, linux-rdma@vger.kernel.org,
	linux-hyperv@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH net-next v2 11/13] net: mana: keep a user-configured RSS table across a queue rebuild
Date: Mon, 10 Aug 2026 23:35:28 -0700	[thread overview]
Message-ID: <20260811063530.2428424-3-longli@microsoft.com> (raw)
In-Reply-To: <20260811063530.2428424-1-longli@microsoft.com>

mana_rss_table_init() overwrites the indirection table with the driver
default every time the queues are rebuilt. Both rebuild paths do it:
mana_alloc_qset() for the ethtool/MTU/XDP queue-set swap, and
mana_alloc_queues() for ndo_open and for the TX-timeout reset.

A table the user installed with "ethtool -X" is therefore lost by
operations that have nothing to do with RSS. Resizing the rings, changing
the MTU, toggling a private flag, attaching an XDP program, or simply
taking the port down and up again all silently reset the steering:

  # ethtool -X ens1 equal 1        # everything to queue 0
  # ethtool -G ens1 rx 1024
  # ethtool -x ens1                # back to 0..15, silently

The entries are queue indices, so they stay meaningful as long as the
queue count does not change, and mana_config_rss() already maps them onto
whichever RX objects the new set has. Carry the table over instead of
regenerating it.

Only a user-configured table is preserved, which netif_is_rxfh_configured()
reports: a driver-generated table must still be rebuilt so that it spreads
over all the queues of the new set. ethtool_check_max_channel() refuses a
channel-count reduction that would leave a user table pointing past the
last queue, so the entries are in range by construction; the bounds check
is a safety net for the rebuild paths that do not come from ethtool, and
reports the table as lost rather than steering to a queue that is gone.

Signed-off-by: Long Li <longli@microsoft.com>
---
 drivers/net/ethernet/microsoft/mana/mana_en.c | 50 ++++++++++++++++++-
 1 file changed, 48 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index 0d87440fbfbee7ac5729945d101eaa0c37745fbe..4cab3f658f2487671d26243d4e91f834580b3c5c 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -3545,6 +3545,38 @@ static void mana_rss_table_init(struct mana_port_context *apc)
 			ethtool_rxfh_indir_default(i, apc->num_queues);
 }
 
+/* Decide whether @apc's indirection table can be carried over to a queue set
+ * with @num_queues queues, instead of being rebuilt from the driver default.
+ *
+ * Only a table the user installed with "ethtool -X" is worth preserving: a
+ * driver-generated one has to be rebuilt so that it spreads over all the
+ * queues the new set actually has.
+ *
+ * ethtool_check_max_channel() already refuses a channel-count reduction that
+ * would leave a user-configured table pointing past the last queue, so the
+ * bounds check below is only a safety net for the rebuild paths that do not
+ * originate from ethtool. If it ever trips, the table cannot be honoured for
+ * the new queue count, so tell the core the user's table is gone rather than
+ * silently steering to queues that no longer exist.
+ */
+static bool mana_rss_table_keep(struct mana_port_context *apc,
+				unsigned int num_queues)
+{
+	u32 i;
+
+	if (!netif_is_rxfh_configured(apc->ndev))
+		return false;
+
+	for (i = 0; i < apc->indir_table_sz; i++) {
+		if (apc->indir_table[i] >= num_queues) {
+			ethtool_rxfh_indir_lost(apc->ndev);
+			return false;
+		}
+	}
+
+	return true;
+}
+
 int mana_disable_vport_rx(struct mana_port_context *apc)
 {
 	return mana_cfg_vport_steering(apc, TRI_STATE_FALSE, false, false,
@@ -3860,7 +3892,12 @@ int mana_alloc_queues(struct net_device *ndev)
 		goto destroy_rxq;
 	}
 
-	mana_rss_table_init(apc);
+	/* Keep a user-configured RSS table across a rebuild; the entries are
+	 * queue indices, so they stay meaningful as long as the queue count
+	 * is unchanged. Only a driver-generated table is regenerated here.
+	 */
+	if (!mana_rss_table_keep(apc, apc->num_queues))
+		mana_rss_table_init(apc);
 
 	err = mana_config_rss(apc, TRI_STATE_TRUE, true, true);
 	if (err) {
@@ -4317,7 +4354,16 @@ int mana_alloc_qset(struct mana_port_context *apc,
 	if (err)
 		goto cleanup_rxq;
 
-	mana_rss_table_init(scratch);
+	/* Carry a user-configured RSS table over to the new set. The entries
+	 * are queue indices, so mana_config_rss() in mana_publish_qset() maps
+	 * them onto the new set's RX objects. A driver-generated table is
+	 * rebuilt instead, so it covers every queue of the new set.
+	 */
+	if (mana_rss_table_keep(apc, num_queues))
+		memcpy(scratch->indir_table, apc->indir_table,
+		       apc->indir_table_sz * sizeof(*apc->indir_table));
+	else
+		mana_rss_table_init(scratch);
 
 	mana_qset_snapshot(scratch, out);
 	return 0;
-- 
2.43.0


  parent reply	other threads:[~2026-08-11  6:35 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-08-11  6:35 [PATCH net-next v2 01/13] net: mana: add queue-set allocation and teardown helpers Long Li
2026-08-11  6:35 ` [PATCH net-next v2 10/13] net: mana: release EQs left idle by a channel-count reduction Long Li
2026-08-11  6:35 ` Long Li [this message]
2026-08-11  6:35 ` [PATCH net-next v2 12/13] net: mana: keep the surviving queues when the channel count is reduced Long Li
2026-08-11  6:35 ` [PATCH net-next v2 13/13] net: mana: keep the existing queues when the channel count is raised Long Li

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=20260811063530.2428424-3-longli@microsoft.com \
    --to=longli@microsoft.com \
    --cc=andrew+netdev@lunn.ch \
    --cc=davem@davemloft.net \
    --cc=decui@microsoft.com \
    --cc=edumazet@google.com \
    --cc=ernis@linux.microsoft.com \
    --cc=haiyangz@microsoft.com \
    --cc=horms@kernel.org \
    --cc=jgg@ziepe.ca \
    --cc=kotaranov@microsoft.com \
    --cc=kuba@kernel.org \
    --cc=kys@microsoft.com \
    --cc=leon@kernel.org \
    --cc=linux-hyperv@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=shradhagupta@linux.microsoft.com \
    --cc=stephen@networkplumber.org \
    --cc=wei.liu@kernel.org \
    /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).