On Sat, Feb 4, 2017 at 9:14 AM, Darren Hart <dvhart@xxxxxxxxxxxxx> wrote:
> On Thu, Jan 26, 2017 at 03:33:01PM +0800, Alex Hung wrote:
>> New firmwares include a feature called 5 button array that supports
>> super key, volume up/down, rotation lock and power button. Especially,
>> support for this feature is required to fix power button on some recent
>> systems.
>>
>> This patch was tested on a Dell Latitude 7280.
>
> Hi Alex,
>
> Minor nit below (no need to resend, but a pair of follow-up cleanups would be
> nice).
>
> Queued to testing.
>
>>
>> Signed-off-by: Alex Hung <alex.hung@xxxxxxxxxxxxx>
>
> Pali, would you care to offer a review or some testing to verify no unexpected
> conflicts with the other dell drivers?
>
>> ---
>> drivers/platform/x86/intel-hid.c | 107 +++++++++++++++++++++++++++++++++++++--
>> 1 file changed, 102 insertions(+), 5 deletions(-)
>>
>> diff --git a/drivers/platform/x86/intel-hid.c b/drivers/platform/x86/intel-hid.c
>> index cb3ab2b..6e796a5 100644
>> --- a/drivers/platform/x86/intel-hid.c
>> +++ b/drivers/platform/x86/intel-hid.c
>> @@ -1,5 +1,5 @@
>> /*
>> - * Intel HID event driver for Windows 8
>> + * Intel HID event & 5 button array driver
>> *
>> * Copyright (C) 2015 Alex Hung <alex.hung@xxxxxxxxxxxxx>
>> * Copyright (C) 2015 Andrew Lutomirski <luto@xxxxxxxxxx>
>> @@ -57,8 +57,24 @@ static const struct key_entry intel_hid_keymap[] = {
>> { KE_END },
>> };
>>
>> +/* 5 button array notification value. */
>> +static const struct key_entry intel_array_keymap[] = {
>> + { KE_KEY, 0xC2, { KEY_LEFTMETA} }, /* Press */
>> + { KE_IGNORE, 0xC3, { KEY_LEFTMETA} }, /* Release */
>> + { KE_KEY, 0xC4, { KEY_VOLUMEUP} }, /* Press */
>> + { KE_IGNORE, 0xC5, { KEY_VOLUMEUP} }, /* Release */
>> + { KE_KEY, 0xC6, { KEY_VOLUMEDOWN} }, /* Press */
>> + { KE_IGNORE, 0xC7, { KEY_VOLUMEDOWN} }, /* Release */
>> + { KE_SW, 0xC8, { .sw = {SW_ROTATE_LOCK, 1} } }, /* Press */
>> + { KE_SW, 0xC9, { .sw = {SW_ROTATE_LOCK, 0} } }, /* Release */
>> + { KE_KEY, 0xCE, { KEY_POWER} }, /* Press */
>> + { KE_IGNORE, 0xCF, { KEY_POWER} }, /* Release */
>> + { KE_END },
>> +};
>> +
>> struct intel_hid_priv {
>> struct input_dev *input_dev;
>> + struct input_dev *array;
>> };
>>
>> static int intel_hid_set_enable(struct device *device, int enable)
>> @@ -78,15 +94,43 @@ static int intel_hid_set_enable(struct device *device, int enable)
>> return 0;
>> }
>>
>> +static void intel_button_array_enable(struct device *device, int enable)
>> +{
>
> As enable is always explicitly passed and is used solely as a boolean value, it
> would preferable for both this and the previous usage above to define it as a
> bool. Being self-consistent is important however, so please consider this for a
> cleanup as a separate patch.
>
>> + struct intel_hid_priv *priv = dev_get_drvdata(device);
>> + acpi_handle handle = ACPI_HANDLE(device);
>> + unsigned long long button_cap;
>> + acpi_status status;
>> +
>> + if (!priv->array)
>> + return;
>> +
>> + /* Query supported platform features */
>> + status = acpi_evaluate_integer(handle, "BTNC", NULL, &button_cap);
>> + if (ACPI_FAILURE(status)) {
>> + dev_warn(device, "failed to get button capability\n");
>> + return;
>> + }
>> +
>> + /* Enable|disable features - power button is always enabled */
>> + status = acpi_execute_simple_method(handle, "BTNE",
>> + enable ? button_cap : 1);
>> + if (ACPI_FAILURE(status))
>> + dev_warn(device, "failed to set button capability\n");
>> +}
>> +
>> static int intel_hid_pl_suspend_handler(struct device *device)
>> {
>> intel_hid_set_enable(device, 0);
>> + intel_button_array_enable(device, 0);
>> +
>> return 0;
>> }
>>
>> static int intel_hid_pl_resume_handler(struct device *device)
>> {
>> intel_hid_set_enable(device, 1);
>> + intel_button_array_enable(device, 1);
>> +
>> return 0;
>> }
>>
>> @@ -126,11 +170,43 @@ static int intel_hid_input_setup(struct platform_device *device)
>> return ret;
>> }
>>
>> +static int intel_button_array_input_setup(struct platform_device *device)
>> +{
>> + struct intel_hid_priv *priv = dev_get_drvdata(&device->dev);
>> + int ret;
>> +
>> + /* Setup input device for 5 button array */
>> + priv->array = input_allocate_device();
>> + if (!priv->array)
>> + return -ENOMEM;
>> +
>> + ret = sparse_keymap_setup(priv->array, intel_array_keymap, NULL);
>> + if (ret)
>> + goto err_free_array_device;
>> +
>> + priv->array->dev.parent = &device->dev;
>> + priv->array->name = "Intel HID 5 button array";
>> + priv->array->id.bustype = BUS_HOST;
>> +
>> + ret = input_register_device(priv->array);
>> + if (ret)
>> + goto err_free_array_device;
>> +
>> + return 0;
>> +
>> +err_free_array_device:
>> + input_free_device(priv->array);
>> + return ret;
>
> This return path is more complex than it could be, since you test for ret before
> return anyway:
>
> out:
> if (ret)
> input_free_device(priv->array);
> return ret;
>
> There is no need for a second return point that I can see. Same for the
> hid_input_setup function. We can remove 8 lines.
>
> This follows the previous nit - it's self consistent, but a follow-on cleanup
> patch would be worthwhile.
>
> Thanks Alex, I've queued this to testing and it will go to for-next unless the
> CI, Pali, or a user reports a problem. Appreciate all your effort on this one.
Thanks for the review. I will send another patch to address all comments later.
>
> --
> Darren Hart
> Intel Open Source Technology Center
--
Cheers,
Alex Hung
