gpio/lynxpoint: check if the interrupt is enabled in IRQ handler

Checking LP_INT_STAT is not enough in the interrupt handler because its
contents get updated regardless of whether the pin has interrupt enabled or
not. This causes the driver to loop forever for GPIOs that are pulled up.

Fix this by checking the interrupt enable bit for the pin as well.

Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

diff --git a/drivers/gpio/gpio-lynxpoint.c b/drivers/gpio/gpio-lynxpoint.c
index 2d9ca6055e5e0bde239247fabe41c960d4dd3020..41b5913ddabe6e0a8b1f417004d86628234b77c5 100644 (file)
--- a/drivers/gpio/gpio-lynxpoint.c
+++ b/drivers/gpio/gpio-lynxpoint.c
@@ -248,14 +248,15 @@ static void lp_gpio_irq_handler(unsigned irq, struct irq_desc *desc)
        struct lp_gpio *lg = irq_data_get_irq_handler_data(data);
        struct irq_chip *chip = irq_data_get_irq_chip(data);
        u32 base, pin, mask;
-       unsigned long reg, pending;
+       unsigned long reg, ena, pending;
        unsigned virq;
 
        /* check from GPIO controller which pin triggered the interrupt */
        for (base = 0; base < lg->chip.ngpio; base += 32) {
                reg = lp_gpio_reg(&lg->chip, base, LP_INT_STAT);
+               ena = lp_gpio_reg(&lg->chip, base, LP_INT_ENABLE);
 
-               while ((pending = inl(reg))) {
+               while ((pending = (inl(reg) & inl(ena)))) {
                        pin = __ffs(pending);
                        mask = BIT(pin);
                        /* Clear before handling so we don't lose an edge */