[SRU][R:linux-azure][PATCH 1/1] net: mana: Cache MANA_QUERY_LINK_CONFIG result to avoid repeated HWC queries
John Cabaj
john.cabaj at canonical.com
Thu Sep 24 03:28:20 UTC 2026
From: Erni Sri Satya Vennela <ernis at linux.microsoft.com>
BugLink: https://bugs.launchpad.net/bugs/2168373
mana_query_link_cfg() sends an HWC command to firmware on every call,
but the link speed and QoS values it returns only change when the
driver explicitly calls mana_set_bw_clamp(). This function is called
not only by userspace via ethtool get_link_ksettings, but also
periodically by hv_netvsc through netvsc_get_link_ksettings and by
the sysfs speed_show attribute via dev_attr_show, resulting in
unnecessary HWC traffic every few minutes.
Add a link_cfg_error field to mana_port_context to cache the query
result. The field uses three states: 1 (not yet queried, initial
value set during mana_probe_port), 0 (success, speed/max_speed are
valid), or a negative errno for permanent errors like -EOPNOTSUPP
when the hardware does not support the command. Transient errors and
qos_unconfigured responses are not cached so that subsequent calls
will retry.
MANA is ops-locked because it implements net_shaper_ops, so the core
already takes netdev_lock() around all ethtool_ops and net_shaper_ops
entry points. Reuse that lock to serialize mana_query_link_cfg() and
mana_set_bw_clamp(). This prevents a concurrent mana_set_bw_clamp()
from racing with an in-flight query and publishing stale pre-clamp
speed/max_speed.
Invalidate the cache inside mana_set_bw_clamp() on success, so all
current and future callers that change the link configuration
automatically trigger a fresh query on the next mana_query_link_cfg()
call. Also reset link_cfg_error during resume in mana_probe() under
netdev_lock(), so that any query already in flight cannot later
store 0 and silently overwrite the post-resume invalidation.
Signed-off-by: Erni Sri Satya Vennela <ernis at linux.microsoft.com>
Link: https://patch.msgid.link/20260606133301.2180073-1-ernis@linux.microsoft.com
Signed-off-by: Jakub Kicinski <kuba at kernel.org>
(cherry picked from commit 8e0ffcc926f42103fdded733a28af16438babc0e)
Signed-off-by: John Cabaj <john.cabaj at canonical.com>
---
drivers/net/ethernet/microsoft/mana/mana_en.c | 23 +++++++++++++++----
include/net/mana/mana.h | 4 ++++
2 files changed, 23 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
index 3239e9de973f..e6fe399e9d17 100644
--- a/drivers/net/ethernet/microsoft/mana/mana_en.c
+++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
@@ -1495,6 +1495,12 @@ int mana_query_link_cfg(struct mana_port_context *apc)
struct mana_query_link_config_req req = {};
int err;
+ netdev_assert_locked(ndev);
+
+ err = apc->link_cfg_error;
+ if (err <= 0)
+ return err;
+
mana_gd_init_req_hdr(&req.hdr, MANA_QUERY_LINK_CONFIG,
sizeof(req), sizeof(resp));
@@ -1507,6 +1513,7 @@ int mana_query_link_cfg(struct mana_port_context *apc)
if (err) {
if (err == -EOPNOTSUPP) {
netdev_info_once(ndev, "MANA_QUERY_LINK_CONFIG not supported\n");
+ apc->link_cfg_error = err;
return err;
}
netdev_err(ndev, "Failed to query link config: %d\n", err);
@@ -1524,12 +1531,12 @@ int mana_query_link_cfg(struct mana_port_context *apc)
return err;
}
- if (resp.qos_unconfigured) {
- err = -EINVAL;
- return err;
- }
+ if (resp.qos_unconfigured)
+ return -EINVAL;
+
apc->speed = resp.link_speed_mbps;
apc->max_speed = resp.qos_speed_mbps;
+ apc->link_cfg_error = 0;
return 0;
}
@@ -1541,6 +1548,8 @@ int mana_set_bw_clamp(struct mana_port_context *apc, u32 speed,
struct net_device *ndev = apc->ndev;
int err;
+ netdev_assert_locked(ndev);
+
mana_gd_init_req_hdr(&req.hdr, MANA_SET_BW_CLAMP,
sizeof(req), sizeof(resp));
req.vport = apc->port_handle;
@@ -1574,6 +1583,8 @@ int mana_set_bw_clamp(struct mana_port_context *apc, u32 speed,
if (resp.qos_unconfigured)
netdev_info(ndev, "QoS is unconfigured\n");
+ /* Invalidate the cache; next query will re-fetch from firmware. */
+ apc->link_cfg_error = 1;
return 0;
}
@@ -3759,6 +3770,7 @@ static int mana_probe_port(struct mana_context *ac, int port_idx,
apc->port_handle = INVALID_MANA_HANDLE;
apc->pf_filter_handle = INVALID_MANA_HANDLE;
apc->port_idx = port_idx;
+ apc->link_cfg_error = 1;
apc->cqe_coalescing_enable = 0;
apc->cqe8_coalescing_enable = 0;
@@ -4084,6 +4096,9 @@ int mana_probe(struct gdma_dev *gd, bool resuming)
rtnl_lock();
apc = netdev_priv(ac->ports[i]);
enable_work(&apc->queue_reset_work);
+ netdev_lock(ac->ports[i]);
+ apc->link_cfg_error = 1;
+ netdev_unlock(ac->ports[i]);
err = mana_attach(ac->ports[i]);
rtnl_unlock();
/* Log the port for which the attach failed, stop
diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
index dbd5922bd81d..278be1ec55f1 100644
--- a/include/net/mana/mana.h
+++ b/include/net/mana/mana.h
@@ -621,6 +621,10 @@ struct mana_port_context {
u32 speed;
/* Maximum speed supported by the SKU (mbps) */
u32 max_speed;
+ /* 1 = not queried, 0 = cached success, negative = permanent error.
+ * Protected by the netdev instance lock.
+ */
+ int link_cfg_error;
bool port_is_up;
bool port_st_save; /* Saved port state */
--
2.43.0
More information about the kernel-team
mailing list