If a phydevice use c45, its phy_id property is always 0, so
this adds a phy_c45_ids sysfs attribute to MDIO devices. This
attribute can be useful when debugging problems related to
phy drivers.
Signed-off-by: Jianhui Zhao <[email protected]>
---
drivers/net/phy/phy_device.c | 24 ++++++++++++++++++++++++
1 file changed, 24 insertions(+)
diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
index 17d0d0555a79..521228792840 100644
--- a/drivers/net/phy/phy_device.c
+++ b/drivers/net/phy/phy_device.c
@@ -560,6 +560,29 @@ phy_id_show(struct device *dev, struct device_attribute *attr, char *buf)
}
static DEVICE_ATTR_RO(phy_id);
+static ssize_t
+phy_c45_ids_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+ struct phy_device *phydev = to_phy_device(dev);
+ const int num_ids = ARRAY_SIZE(phydev->c45_ids.device_ids);
+ int ret = 0;
+ int i;
+
+ if (!phydev->is_c45)
+ return 0;
+
+ for (i = 1; i < num_ids; i++) {
+ if (phydev->c45_ids.device_ids[i] == 0xffffffff)
+ continue;
+
+ ret += sprintf(buf + ret, "%d: 0x%.8lx\n", i,
+ (unsigned long)phydev->c45_ids.device_ids[i]);
+ }
+
+ return ret;
+}
+static DEVICE_ATTR_RO(phy_c45_ids);
+
static ssize_t
phy_interface_show(struct device *dev, struct device_attribute *attr, char *buf)
{
@@ -597,6 +620,7 @@ static DEVICE_ATTR_RO(phy_dev_flags);
static struct attribute *phy_dev_attrs[] = {
&dev_attr_phy_id.attr,
+ &dev_attr_phy_c45_ids.attr,
&dev_attr_phy_interface.attr,
&dev_attr_phy_has_fixups.attr,
&dev_attr_phy_dev_flags.attr,
--
2.34.1
diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
index 17d0d0555a79..81a4bc043ee2 100644
--- a/drivers/net/phy/phy_device.c
+++ b/drivers/net/phy/phy_device.c
@@ -541,8 +541,10 @@ static int phy_bus_match(struct device *dev, struct device_driver *drv)
if ((phydrv->phy_id & phydrv->phy_id_mask) ==
(phydev->c45_ids.device_ids[i] &
- phydrv->phy_id_mask))
+ phydrv->phy_id_mask)) {
+ phydev->phy_id = phydev->c45_ids.device_ids[i];
return 1;
+ }
}
return 0;
} else {
How about modifying it like this?
On Mon, Jun 12, 2023 at 10:04:26PM +0800, Jianhui Zhao wrote:
> diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
> index 17d0d0555a79..81a4bc043ee2 100644
> --- a/drivers/net/phy/phy_device.c
> +++ b/drivers/net/phy/phy_device.c
> @@ -541,8 +541,10 @@ static int phy_bus_match(struct device *dev, struct device_driver *drv)
>
> if ((phydrv->phy_id & phydrv->phy_id_mask) ==
> (phydev->c45_ids.device_ids[i] &
> - phydrv->phy_id_mask))
> + phydrv->phy_id_mask)) {
> + phydev->phy_id = phydev->c45_ids.device_ids[i];
> return 1;
> + }
> }
> return 0;
> } else {
>
> How about modifying it like this?
O.K.
Please update the sysfs documentation as well.
Andrew
On Mon, Jun 12, 2023 at 10:04:26PM +0800, Jianhui Zhao wrote:
> diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
> index 17d0d0555a79..81a4bc043ee2 100644
> --- a/drivers/net/phy/phy_device.c
> +++ b/drivers/net/phy/phy_device.c
> @@ -541,8 +541,10 @@ static int phy_bus_match(struct device *dev, struct device_driver *drv)
>
> if ((phydrv->phy_id & phydrv->phy_id_mask) ==
> (phydev->c45_ids.device_ids[i] &
> - phydrv->phy_id_mask))
> + phydrv->phy_id_mask)) {
> + phydev->phy_id = phydev->c45_ids.device_ids[i];
> return 1;
> + }
> }
> return 0;
> } else {
>
> How about modifying it like this?
No - there are C45 PHYs where the ID in each MMD are different. 88x3310
is one such example. If we're going to report any of them, we should
report all of them.
--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
On Mon, Jun 12, 2023 at 04:06:32PM +0100, Russell King (Oracle) wrote:
> On Mon, Jun 12, 2023 at 10:04:26PM +0800, Jianhui Zhao wrote:
> > diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
> > index 17d0d0555a79..81a4bc043ee2 100644
> > --- a/drivers/net/phy/phy_device.c
> > +++ b/drivers/net/phy/phy_device.c
> > @@ -541,8 +541,10 @@ static int phy_bus_match(struct device *dev, struct device_driver *drv)
> >
> > if ((phydrv->phy_id & phydrv->phy_id_mask) ==
> > (phydev->c45_ids.device_ids[i] &
> > - phydrv->phy_id_mask))
> > + phydrv->phy_id_mask)) {
> > + phydev->phy_id = phydev->c45_ids.device_ids[i];
> > return 1;
> > + }
> > }
> > return 0;
> > } else {
> >
> > How about modifying it like this?
>
> No - there are C45 PHYs where the ID in each MMD are different. 88x3310
> is one such example. If we're going to report any of them, we should
> report all of them.
The other issue is, this will remain as zero until a driver has
successfully matched the device, which means as far as userspace
knowing what type of PHY it is (to find a driver) this doesn't
improve the situation.
I don't think this is a very good solution.
--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!