Skip to content

Commit

Permalink
net: omit ndo_hwtstamp_get() call when possible in dev_set_hwtstamp_p…
Browse files Browse the repository at this point in the history
…hylib()

Setting dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS is only legal
for drivers which were converted to ndo_hwtstamp_get() and
ndo_hwtstamp_set(), and it is only there that we call ndo_hwtstamp_set()
for a request that otherwise goes to phylib (for stuff like packet traps,
which need to be undone if phylib failed, hence the old_cfg logic).

The problem is that we end up calling ndo_hwtstamp_get() when we don't
need to (even if the SIOCSHWTSTAMP wasn't intended for phylib, or if it
was, but the driver didn't set IFF_SEE_ALL_HWTSTAMP_REQUESTS). For those
unnecessary conditions, we share a code path with virtual drivers (vlan,
macvlan, bonding) where ndo_hwtstamp_get() is implemented as
generic_hwtstamp_get_lower(), and may be resolved through
generic_hwtstamp_ioctl_lower() if the lower device is unconverted.

I.e. this situation:

$ ip link add link eno0 name eno0.100 type vlan id 100
$ hwstamp_ctl -i eno0.100 -t 1

We are unprepared to deal with this, because if ndo_hwtstamp_get() is
resolved through a legacy ndo_eth_ioctl(SIOCGHWTSTAMP) lower_dev
implementation, that needs a non-NULL old_cfg.ifr pointer, and we don't
have it.

But we don't even need to deal with it either. In the general case,
drivers may not even implement SIOCGHWTSTAMP handling, only SIOCSHWTSTAMP,
so it makes sense to completely avoid a SIOCGHWTSTAMP call if we can.

The solution is to split the single "if" condition into 3 smaller ones,
thus separating the decision to call ndo_hwtstamp_get() from the
decision to call ndo_hwtstamp_set(). The third "if" condition is
identical to the first one, and both are subsets of the second one.
Thus, the "cfg" argument of kernel_hwtstamp_config_changed() is always
valid.

Reported-by: Eric Dumazet <edumazet@google.com>
Closes: https://lore.kernel.org/netdev/CANn89iLOspJsvjPj+y8jikg7erXDomWe8sqHMdfL_2LQSFrPAg@mail.gmail.com/
Fixes: fd770e8 ("net: remove phy_has_hwtstamp() -> phy_mii_ioctl() decision from converted drivers")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
  • Loading branch information
Vladimir Oltean authored and David S. Miller committed Aug 6, 2023
1 parent 54024db commit c35e927
Showing 1 changed file with 5 additions and 2 deletions.
7 changes: 5 additions & 2 deletions net/core/dev_ioctl.c
Original file line number Diff line number Diff line change
Expand Up @@ -334,20 +334,23 @@ static int dev_set_hwtstamp_phylib(struct net_device *dev,

cfg->source = phy_ts ? HWTSTAMP_SOURCE_PHYLIB : HWTSTAMP_SOURCE_NETDEV;

if (!phy_ts || (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS)) {
if (phy_ts && (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS)) {
err = ops->ndo_hwtstamp_get(dev, &old_cfg);
if (err)
return err;
}

if (!phy_ts || (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS)) {
err = ops->ndo_hwtstamp_set(dev, cfg, extack);
if (err) {
if (extack->_msg)
netdev_err(dev, "%s\n", extack->_msg);
return err;
}
}

if (phy_ts && (dev->priv_flags & IFF_SEE_ALL_HWTSTAMP_REQUESTS))
changed = kernel_hwtstamp_config_changed(&old_cfg, cfg);
}

if (phy_ts) {
err = phy_hwtstamp_set(dev->phydev, cfg, extack);
Expand Down

0 comments on commit c35e927

Please sign in to comment.