2013/6/20 Alexey Brodkin <[email protected]>: > On 06/20/2013 01:57 PM, Florian Fainelli wrote: >> 2013/6/20 Alexey Brodkin <[email protected]>: >> [snip] >> >>> >>> In general MDIO register gets polled by "libphy" once in a couple of >>> seconds, so delay of 25 milliseconds IMHO is fine. >>> >>>>> +int arc_mdio_probe(struct device_node *dev_node, struct arc_emac_priv >>>>> *priv) >>>>> +{ >>>> >>>>> + snprintf(bus->id, MII_BUS_ID_SIZE, "%.8x", (unsigned >>>>> int)priv->regs); >>>> >>>> Is bus->id exposed to user-space somehow? >>> >>> Well as a boot-up message from "libphy": >>> ==== >>> libphy: Synopsys MII Bus: probed >>> ==== >> >> Well not only as a boot-up message, this serves as unique identifer in >> the entire system for your MDIO bus. It is crucial that you have an >> unique MDIO bus identifier for at least the two following reasons: >> >> - the corresponding kobject/sysfs node that is going to be created >> also needs to be unique in the system >> - you may have multiple MDIO bus in the system (e.g: the one for your >> specific driver and the fixed MDIO bus) > > Good explanation. I didn't realize it yet. > And it seems like I talked about bus "name", not "id". > >> Since you are using Device Tree already, you may just turn this into: >> >> snprintf(bus->id, MII_BUS_ID_SIZE, "%s", pdev->name); >> >> assuming that your node is already properly labelleled (e.g: >> mdio@deadbeef) which would be exactly equivalent to what you are doing >> with priv->regs; > > The thing is I don't have separate "mdio" device description in DT. > Regs value is read from "ethernet" device description. > > Here's my DT entry for entire emac: > ============= > ethernet@c0fc2000 { > compatible = "snps,arc-emac"; > reg = <0xc0fc2000 0x3c>; > interrupts = <6>; > mac-address = [ 00 11 22 33 44 55 ]; > clock-frequency = <80000000>; > max-speed = <100>; > phy = <&phy0>; > > #address-cells = <1>; > #size-cells = <0>; > phy0: ethernet-phy@0 { > reg = <1>; > }; > }; > ============= > > So with your proposal bus id will be just "ethernet" which is not that > unique, right?
No, it would be c0fc2000, which is, so we are just fine. Drivers that support DT instantiation or platform_device/driver should in general use <pdev->name>-<pdev->id> which works fine in both cases. -- Florian -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
