[U-Boot] [PATCH 09/10] exynos: Enable the LCD backlight for snow

[Simon Glass]

Hi Ajay,

On 29 March 2014 16:35, Ajay kumar <ajaynumb at gmail.com> wrote:

> Hi Simon,
>
> Find my comments below:
>
> On Wed, Mar 26, 2014 at 11:26 PM, Simon Glass <sjg at chromium.org> wrote:
>
>> The backlight uses FETs on the TPS65090. Enable this so that the display
>> is visible.
>>
>> Signed-off-by: Simon Glass <sjg at chromium.org>
>> ---
>>
>>  board/samsung/smdk5250/exynos5-dt.c | 90
>> +++++++++++++++++++++++++++++++++++++
>>  1 file changed, 90 insertions(+)
>>
>> diff --git a/board/samsung/smdk5250/exynos5-dt.c
>> b/board/samsung/smdk5250/exynos5-dt.c
>> index c83b034..a0ae641 100644
>> --- a/board/samsung/smdk5250/exynos5-dt.c
>> +++ b/board/samsung/smdk5250/exynos5-dt.c
>> @@ -20,6 +20,7 @@
>>  #include <asm/arch/sromc.h>
>>  #include <power/pmic.h>
>>  #include <power/max77686_pmic.h>
>> +#include <power/tps65090_pmic.h>
>>  #include <tmu.h>
>>
>>  DECLARE_GLOBAL_DATA_PTR;
>> @@ -60,6 +61,52 @@ int checkboard(void)
>>  #endif
>>
>>  #ifdef CONFIG_LCD
>>
> We should make sure this code runs only for snow, and not for SMDK.
> Only Snow contains the bridge chip.
>

I'll add it to the device tree.


> +static int board_dp_bridge_setup(void)
>> +{
>> +       struct exynos5_gpio_part1 *gpio1 =
>> +               (struct exynos5_gpio_part1
>> *)samsung_get_base_gpio_part1();
>> +       const int MAX_TRIES = 10;
>> +       int num_tries;
>> +
>> +       debug("%s\n", __func__);
>> +
>> +       /* Setup the GPIOs */
>> +
>> +       /* PD is ACTIVE_LOW, and initially de-asserted */
>> +       s5p_gpio_set_pull(&gpio1->y2, 5, GPIO_PULL_NONE);
>> +       s5p_gpio_direction_output(&gpio1->y2, 5, 1);
>> +
>> +       /* Reset is ACTIVE_LOW */
>> +       s5p_gpio_set_pull(&gpio1->x1, 5, GPIO_PULL_NONE);
>> +       s5p_gpio_direction_output(&gpio1->x1, 5, 0);
>> +
>> +       udelay(10);
>> +       s5p_gpio_set_value(&gpio1->x1, 5, 1);
>> +
>> +       s5p_gpio_direction_input(&gpio1->x0, 7);
>> +
>> +       /*
>> +        * We need to wait for 90ms after bringing up the bridge since
>> there
>> +        * is a phantom "high" on the HPD chip during its bootup.  The
>> phantom
>> +        * high comes within 7ms of de-asserting PD and persists for at
>> least
>> +        * 15ms.  The real high comes roughly 50ms after PD is
>> de-asserted. The
>> +        * phantom high makes it hard for us to know when the NXP chip is
>> up.
>> +        */
>> +       mdelay(90);
>> +
>> +       for (num_tries = 0; num_tries < MAX_TRIES; num_tries++) {
>> +               /* Check HPD.  If it's high, we're all good. */
>> +               if (s5p_gpio_get_value(&gpio1->x0, 7))
>> +                               return 0;
>> +
>> +               debug("%s: eDP bridge failed to come up; try %d of %d\n",
>> +                     __func__, num_tries, MAX_TRIES);
>> +       }
>> +
>> +       /* Immediately go into bridge reset if the hp line is not high */
>> +       return -ENODEV;
>> +}
>> +
>>  void exynos_cfg_lcd_gpio(void)
>>  {
>>         struct exynos5_gpio_part1 *gpio1 =
>> @@ -81,4 +128,47 @@ void exynos_set_dp_phy(unsigned int onoff)
>>  {
>>         set_dp_phy_ctrl(onoff);
>>  }
>> +
>> +void exynos_backlight_on(unsigned int onoff)
>> +{
>> +       debug("%s(%u)\n", __func__, onoff);
>> +
>> +       if (onoff) {
>> +#ifdef CONFIG_POWER_TPS65090
>> +               struct exynos5_gpio_part1 *gpio1 =
>> +                       (struct exynos5_gpio_part1 *)
>> +                                       samsung_get_base_gpio_part1();
>> +               int ret;
>> +
>> +               ret = tps65090_fet_enable(1); /* Enable FET1, backlight */
>> +               if (ret)
>> +                       return;
>> +
>> +               /* T5 in the LCD timing spec (defined as > 10ms) */
>> +               mdelay(10);
>>
> Can you check for backlight without using the above delay?
>

I think we need to keep these delays due to the LCD timing spec. It often
does work, but we need it to always work.


> +
>> +               /* board_dp_backlight_pwm */
>> +               s5p_gpio_direction_output(&gpio1->b2, 0, 1);
>>
> +
>> +               /* T6 in the LCD timing spec (defined as > 10ms) */
>> +               mdelay(10);
>>
> Can you check for backlight without using the above delay?
>
>> +
>> +               /* board_dp_backlight_en */
>> +               s5p_gpio_direction_output(&gpio1->x3, 0, 1);
>> +#endif
>> +       }
>> +}
>> +
>> +void exynos_lcd_power_on(void)
>> +{
>> +       debug("%s\n", __func__);
>> +
>> +#ifdef CONFIG_POWER_TPS65090
>> +       /* board_dp_lcd_vdd */
>> +       tps65090_fet_enable(6); /* Enable FET6, lcd panel */
>> +#endif
>> +
>> +       board_dp_bridge_setup();
>>
> returning int. check for return value and print in case of failure.
>

Will do.

Regads,
Simon