Hi, On Mon, Aug 01, 2016 at 09:39:34PM +0800, Chen-Yu Tsai wrote: > Hi, > > On Sat, Jul 30, 2016 at 11:20 PM, maxime.ripard@free-electrons.com > wrote: > > On Sat, Jul 30, 2016 at 10:52:45PM +0800, Icenowy Zheng wrote: > >> > + if (of_device_is_compatible(pdev->dev.of_node, > >> > + "allwinner,sun6i-a31-spdif")) { > >> > + host->rst = devm_reset_control_get_optional(&pdev->dev, NULL); > >> > + if (IS_ERR(host->rst) && PTR_ERR(host->rst) == -EPROBE_DEFER) { > >> > + ret = -EPROBE_DEFER; > >> > + dev_err(&pdev->dev, "Failed to get reset: %d\n", ret); > >> > + goto err_disable_apb_clk; > >> > + } > >> > + if (!IS_ERR(host->rst)) > >> > + reset_control_deassert(host->rst); > >> > + } > >> > + > >> I think you do not need the compatible. > >> You can just detect whether the reset is present. > > > > That would weaken the error check. If we're running on the A31 and are > > missing our reset property, it would go unnoticed. > > We've been doing it this way with the mmc controller and the usb hosts though. > IIRC you once said in the older SoCs, the reset control is tied to the clock > gate in the hardware. > > The _optional variant is also funny, though I understand it is a design > of the reset controller framework. Yes, I know. But that doesn't prevent that design from being better. Maxime -- Maxime Ripard, Free Electrons Embedded Linux and Kernel engineering http://free-electrons.com