Re: [PATCH 1/3] net: macb: fix for fixed-link mode

From: Russell King - ARM Linux admin
Date: Tue Dec 10 2019 - 06:38:53 EST


On Tue, Dec 10, 2019 at 09:14:13AM +0000, Milind Parab wrote:
> >> This patch fix the issue with fixed link. With fixed-link
> >> device opening fails due to macb_phylink_connect not
> >> handling fixed-link mode, in which case no MAC-PHY connection
> >> is needed and phylink_connect return success (0), however
> >> in current driver attempt is made to search and connect to
> >> PHY even for fixed-link.
> >>
> >> Signed-off-by: Milind Parab <mparab@xxxxxxxxxxx>
> >> ---
> >> drivers/net/ethernet/cadence/macb_main.c | 17 ++++++++---------
> >> 1 file changed, 8 insertions(+), 9 deletions(-)
> >>
> >> diff --git a/drivers/net/ethernet/cadence/macb_main.c b/drivers/net/ethernet/cadence/macb_main.c
> >> index 9c767ee252ac..6b68ef34ab19 100644
> >> --- a/drivers/net/ethernet/cadence/macb_main.c
> >> +++ b/drivers/net/ethernet/cadence/macb_main.c
> >> @@ -615,17 +615,13 @@ static int macb_phylink_connect(struct macb *bp)
> >> {
> >> struct net_device *dev = bp->dev;
> >> struct phy_device *phydev;
> >> + struct device_node *dn = bp->pdev->dev.of_node;
> >> int ret;
> >>
> >> - if (bp->pdev->dev.of_node &&
> >> - of_parse_phandle(bp->pdev->dev.of_node, "phy-handle", 0)) {
> >> - ret = phylink_of_phy_connect(bp->phylink, bp->pdev-
> >>dev.of_node,
> >> - 0);
> >> - if (ret) {
> >> - netdev_err(dev, "Could not attach PHY (%d)\n", ret);
> >> - return ret;
> >> - }
> >> - } else {
> >> + if (dn)
> >> + ret = phylink_of_phy_connect(bp->phylink, dn, 0);
> >> +
> >> + if (!dn || (ret && !of_parse_phandle(dn, "phy-handle", 0))) {
> >
> >Hi,
> >If of_parse_phandle() returns non-null, the device_node it returns will
> >have its reference count increased by one. That reference needs to be
> >put.
> >
>
> Okay, as per your suggestion below addition will be okay to store the "phy_node" and then of_node_put(phy_node) on error
>
> phy_node = of_parse_phandle(dn, "phy-handle", 0);
> if (!dn || (ret && !phy_node)) {
> phydev = phy_find_first(bp->mii_bus);
...
> if (phy_node)
> of_node_put(phy_node);

As you're only interested in whether phy-handle exists or not, you
could do this instead:

phy_node = of_parse_phandle(dn, "phy-handle", 0);
of_node_put(phy_node);
if (!dn || (ret && !phy_node)) {
...

Yes, it looks a bit weird, but the only thing you're interested in
here is whether of_parse_phandle() returned NULL or non-NULL. You're
not interested in dereferencing the pointer.

Some may raise some eye-brows at that, so it may be better to have
this as a helper:

static bool macb_phy_handle_exists(struct device_node *dn)
{
dn = of_parse_phandle(dn, "phy-handle", 0);
of_node_put(dn);
return dn != NULL;
}

and use it as:

if (!dn || (ret && !macb_phy_handle_exists(dn))) {

which is more obvious what is going on.

>
> return ret;
>
> >I assume you're trying to determine whether phylink_of_phy_connect()
> >failed because of a missing phy-handle rather than of_phy_attach()
> >failing? Maybe those two failures ought to be distinguished by errno
> >return value?
>
> Yes, PHY will be scanned only if phylink_of_phy_connect() returns error due to missing "phy-handle".
> Currently, phylink_of_phy_connect() returns same error for missing "phy-handle" and of_phy_attach() failure.
>
> >of_phy_attach() may fail due to of_phy_find_device() failing to find
> >the PHY, or phy_attach_direct() failing. We could switch from using
> >of_phy_attach(), to using of_phy_find_device() directly so we can then
> >propagate phy_attach_direct()'s error code back, rather than losing it.
> >That would then leave the case of of_phy_find_device() failure to be
> >considered in terms of errno return value.

Here's a patch I quickly knocked up that does this - may not apply to
the kernel you're using as there's a whole bunch of work I have
outstanding, but gives the outline idea. Does this help?

8<===
From: Russell King <rmk+kernel@xxxxxxxxxxxxxxx>
Subject: [PATCH] net: phylink: avoid of_phy_attach()

of_phy_attach() hides the return value of phy_attach_direct(), forcing
us to return a "generic" ENODEV error code that is indistinguishable
from the lack-of-phy-property case.

Switch to using of_phy_find_device() to find the PHY device, and then
propagating any phy_attach_direct() error back to the caller.

Signed-off-by: Russell King <rmk+kernel@xxxxxxxxxxxxxxx>
---
drivers/net/phy/phylink.c | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c
index e9036b72114c..5a5109428d9e 100644
--- a/drivers/net/phy/phylink.c
+++ b/drivers/net/phy/phylink.c
@@ -887,14 +887,17 @@ int phylink_of_phy_connect(struct phylink *pl, struct device_node *dn,
return 0;
}

- phy_dev = of_phy_attach(pl->netdev, phy_node, flags,
- pl->link_interface);
+ phy_dev = of_phy_find_device(phy_node);
/* We're done with the phy_node handle */
of_node_put(phy_node);
-
if (!phy_dev)
return -ENODEV;

+ ret = phy_attach_direct(pl->netdev, phy_dev, flags,
+ pl->link_interface);
+ if (ret)
+ return ret;
+
ret = phylink_bringup_phy(pl, phy_dev, pl->link_config.interface);
if (ret)
phy_detach(phy_dev);
--
2.20.1


--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up
According to speedtest.net: 11.9Mbps down 500kbps up