| Message ID | 20230315174016.4193015-1-joshwash@google.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] gve: Cache link_speed value from device | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Clearly marked for net |
| netdev/apply | fail | Patch does not apply to net |
On Wed, 15 Mar 2023 10:40:16 -0700 joshwash@google.com wrote: > From: Joshua Washington <joshwash@google.com> > > The link speed is never changed for the uptime of a VM. Caching the > value will allow for better performance. > > Fixes: 7e074d5a76ca ("gve: Enable Link Speed Reporting in the driver.") If it needs to go in as a fix / to stable we need a bit more info about the nature of the problem. What user-visible (or hypervisor-visible) impact will be? What entity needs link info so often that it becomes a problem? The code looks fine as an optimization, i.e. for net-next, but you say "PATCH net" and there's a Fixes tag...
On Thu, 16 Mar 2023 17:42:27 -0700 Jakub Kicinski wrote: > If it needs to go in as a fix / to stable we need a bit more info about > the nature of the problem. What user-visible (or hypervisor-visible) > impact will be? What entity needs link info so often that it becomes > a problem? > > The code looks fine as an optimization, i.e. for net-next, but you say > "PATCH net" and there's a Fixes tag... Hm, actually it doesn't apply to either tree..
diff --git a/google/gve/gve_ethtool.c b/google/gve/gve_ethtool.c index b18804e..cfd4b8d 100644 --- a/google/gve/gve_ethtool.c +++ b/google/gve/gve_ethtool.c @@ -584,7 +584,10 @@ static int gve_get_link_ksettings(struct net_device *netdev, struct ethtool_link_ksettings *cmd) { struct gve_priv *priv = netdev_priv(netdev); - int err = gve_adminq_report_link_speed(priv); + int err = 0; + + if (priv->link_speed == 0) + err = gve_adminq_report_link_speed(priv); cmd->base.speed = priv->link_speed; return err;