Hi Meng,
On Fri, 22 Apr 2016 09:58:34 +0000 Meng Yi meng.yi@nxp.com wrote:
Hi Boris,
-----Original Message----- From: Boris Brezillon [mailto:boris.brezillon@free-electrons.com]
+static int sii902x_bridge_attach(struct drm_bridge *bridge) {
- const struct drm_connector_funcs *funcs = &sii902x_connector_funcs;
- struct sii902x *sii902x = bridge_to_sii902x(bridge);
- struct drm_device *drm = bridge->dev;
- int ret;
- drm_connector_helper_add(&sii902x->connector,
&sii902x_connector_helper_funcs);- if (drm_core_check_feature(drm, DRIVER_ATOMIC))
funcs = &sii902x_atomic_connector_funcs;- ret = drm_connector_init(drm, &sii902x->connector, funcs,
DRM_MODE_CONNECTOR_HDMIA);- if (ret)
return ret;I think drm_connector_register should be used here, or drmlib can't find this connector.
Hm, I think this should be left to the DRM master device (some of them already call drm_connector_register_all() at the end of their ->probe() function).
- if (sii902x->i2c->irq > 0)
sii902x->connector.polled = DRM_CONNECTOR_POLL_HPD;- else
sii902x->connector.polled =DRM_CONNECTOR_POLL_CONNECT;
- drm_mode_connector_attach_encoder(&sii902x->connector,
+bridge->encoder);
- return 0;
+}
...
+static int sii902x_probe(struct i2c_client *client,
const struct i2c_device_id *id)+{
- struct device *dev = &client->dev;
- unsigned int status = 0;
- struct sii902x *sii902x;
- u8 chipid[4];
- int ret;
- sii902x = devm_kzalloc(dev, sizeof(*sii902x), GFP_KERNEL);
- if (!sii902x)
return -ENOMEM;- sii902x->i2c = client;
- sii902x->regmap = devm_regmap_init_i2c(client,
&sii902x_regmap_config);
- if (IS_ERR(sii902x->regmap))
return PTR_ERR(sii902x->regmap);- sii902x->reset_gpio = devm_gpiod_get(dev, "reset",
GPIOD_OUT_LOW);
- if (IS_ERR(sii902x->reset_gpio)) {
dev_err(dev, "Failed to retrieve/request reset gpio: %ld\n",PTR_ERR(sii902x->reset_gpio));return PTR_ERR(sii902x->reset_gpio);Maybe we can use "warning" instead of "dev_err" if there is no "reset_gpio" , because some board using CPLD to manage the reset order , they can't reset the HDMI chip independently.
Sure, since it's optional maybe we should use dev_info().
Thanks for your review.
Regards,
Boris