On 24/03/2025 19:48, Marek Vasut wrote: > On 3/24/25 10:57 AM, Paul Barker wrote: >> On 24/03/2025 01:08, Marek Vasut wrote: >>> On 3/19/25 1:03 PM, Paul Barker wrote: >>> >>> [...] >>> >>>> static int ravb_probe(struct udevice *dev) >>>> { >>>> + struct ravb_device_ops *device_ops = >>>> + (struct ravb_device_ops *)dev_get_driver_data(dev); >>>> struct eth_pdata *pdata = dev_get_plat(dev); >>>> struct ravb_priv *eth = dev_get_priv(dev); >>>> struct mii_dev *mdiodev; >>>> @@ -648,16 +653,32 @@ static int ravb_probe(struct udevice *dev) >>>> if (ret) >>>> goto err_clk_enable; >>>> >>>> + if (device_ops->has_reset) { >>>> + ret = reset_get_by_index(dev, 0, ð->rst); >>> >>> Can devm_reset_control_get_optional() maybe be used here instead of the >>> flag ? >> >> The Ethernet device tree nodes for RZ/G2{H,M,N,E} also contain a reset, >> so we can't use the _optional() function here to perform the reset only >> for RZ/G2L. > Does it pose a problem if we trigger a reset on RZ/G2(non-L) ?
Hi Marek, I don't have the resources to easily answer this question - we would have to consider RZ/G1x and R-Car SoCs which use the RAVB driver as well as the RZ/G2H family. My aim has been to avoid any change in behaviour for non-RZ/G2L platforms which use the RAVB driver. Thanks, -- Paul Barker
OpenPGP_0x27F4B3459F002257.asc
Description: OpenPGP public key
OpenPGP_signature.asc
Description: OpenPGP digital signature