Kever,
On 26 Feb 2018, at 04:07, Kever Yang kever.yang@rock-chips.com wrote:
Hi Simon,
On 02/25/2018 11:54 PM, Simon Glass wrote:
Hi Kever,
On 23 February 2018 at 19:08, Kever Yang kever.yang@rock-chips.com wrote:
Hi Simon,
On 02/12/2018 10:35 PM, Simon Glass wrote:
Hi Kever,
On 8 February 2018 at 19:56, Kever Yang kever.yang@rock-chips.com wrote:
Add api for who can not get phandle from a device property.
Can you please add a motivation to the commit message? It is not obvious to me when this function would be used.
Here is the example why I need this, see the dts node here: lvds@ff2e0000 { ... rockchip,grf = <&grf>; port { port@0 { endpoint@0 { remote-endpoint = <&vopl_out_lvds>; } } } }
We can only get 'grf' udevice by uclass_get_device_by_phandle(), but we not able to get udevice 'vopl_out_lvds', other driver like rockchip pinctrl also need to get udevice by a phandle which is not one of direct property of another device node.
OK I see. This sort of info is useful in the commit message. It helps to know two things about a patch:
- why it is needed
- what it does
In this case see rk_display_init() for how it handles the 'remote-endpoint' property. I think it would be better to have something like:
ofnode ofnode_lookup_phandle(ofnode node, const char *prop_name);
Yes, this is enough for "remote-endpoint" case, but not for pinctrl case: 1199 pinctrl: pinctrl { 2095 pwm0 { 2096 pwm0_pin: pwm0-pin { 2097 rockchip,pins = 2098 <0 RK_PB7 RK_FUNC_1 &pcfg_pull_none>; 2099 }; 2100 };
I think ofnode_lookup_phandle() still not able to find node for 'pcfg_pull_none’.
I had to convert rk_vop to livetree last week to fix a regression on HDMI for our RK3399-Q7. In doing so, I created some code to “walk the tree to find a parent with a given uclass) … this had me introduce an new ofnode_get_parent() function.
Anatolij merged this on the weekend and it’s on Tom’s master already.
Best, Philipp.
Thanks,
- Kever
This should meet your needs without needing to decoding the phandle property in the caller.
Regards, Simon