@@ -376,14 +376,15 @@ void omap_sram_idle(void)
core_next_state = pwrdm_read_next_pwrst(core_pwrdm);
if (per_next_state < PWRDM_POWER_ON) {
omap_uart_prepare_idle(2);
- omap2_gpio_prepare_for_retention();
if (per_next_state == PWRDM_POWER_OFF) {
if (core_next_state == PWRDM_POWER_ON) {
per_next_state = PWRDM_POWER_RET;
pwrdm_set_next_pwrst(per_pwrdm, per_next_state);
per_state_modified = 1;
- } else
+ } else {
+ omap2_gpio_prepare_for_retention();
omap3_per_save_context();
+ }
}
}
@@ -454,9 +455,10 @@ void omap_sram_idle(void)
/* PER */
if (per_next_state < PWRDM_POWER_ON) {
per_prev_state = pwrdm_read_prev_pwrst(per_pwrdm);
- if (per_prev_state == PWRDM_POWER_OFF)
+ if (per_prev_state == PWRDM_POWER_OFF) {
omap3_per_restore_context();
- omap2_gpio_resume_after_retention();
+ omap2_gpio_resume_after_retention();
+ }
omap_uart_resume_idle(2);
if (per_state_modified)
pwrdm_set_next_pwrst(per_pwrdm, PWRDM_POWER_OFF);
@@ -731,7 +731,9 @@ static inline void set_24xx_gpio_triggering(struct gpio_bank *bank, int gpio,
__raw_writel(1 << gpio, bank->base
+ OMAP24XX_GPIO_CLEARWKUENA);
}
- } else {
+ }
+ /* This part needs to be executed always for OMAP34xx */
+ if (cpu_is_omap34xx() || (bank->non_wakeup_gpios & gpio_bit)) {
if (trigger != 0)
bank->enabled_non_wakeup_gpios |= gpio_bit;
else
@@ -1845,7 +1847,8 @@ static int __init _omap_gpio_init(void)
__raw_writel(0, bank->base +
OMAP24XX_GPIO_CTRL);
}
- if (i < ARRAY_SIZE(non_wakeup_gpios))
+ if (cpu_is_omap24xx() &&
+ i < ARRAY_SIZE(non_wakeup_gpios))
bank->non_wakeup_gpios = non_wakeup_gpios[i];
gpio_count = 32;
}
@@ -2031,10 +2034,13 @@ static int workaround_enabled;
void omap2_gpio_prepare_for_retention(void)
{
int i, c = 0;
+ int min = 0;
+ if (cpu_is_omap34xx())
+ min = 1;
/* Remove triggering for all non-wakeup GPIOs. Otherwise spurious
* IRQs will be generated. See OMAP2420 Errata item 1.101. */
- for (i = 0; i < gpio_bank_count; i++) {
+ for (i = min; i < gpio_bank_count; i++) {
struct gpio_bank *bank = &gpio_bank[i];
u32 l1, l2;
@@ -2088,10 +2094,13 @@ void omap2_gpio_prepare_for_retention(void)
void omap2_gpio_resume_after_retention(void)
{
int i;
+ int min = 0;
if (!workaround_enabled)
return;
- for (i = 0; i < gpio_bank_count; i++) {
+ if (cpu_is_omap34xx())
+ min = 1;
+ for (i = min; i < gpio_bank_count; i++) {
struct gpio_bank *bank = &gpio_bank[i];
u32 l, gen, gen0, gen1;
@@ -2119,7 +2128,7 @@ void omap2_gpio_resume_after_retention(void)
* horribly racy, but it's the best we can do to work around
* this silicon bug. */
l ^= bank->saved_datain;
- l &= bank->non_wakeup_gpios;
+ l &= bank->enabled_non_wakeup_gpios;
/*
* No need to generate IRQs for the rising edge for gpio IRQs