[PATCH 0/3] RFC: fix bug http://docs.openmoko.org/trac/ticket/1884

[Werner Almesberger]

Andy Green wrote:
> It'd be neat if this was behind the other funnies with level interrupts
> seen earlier in the year.

I have some vague recollection that we had some issues there, but
nothing specific pops up in my mind. Do you remember some places
worth looking at ?

By the way, I also tried this with the SDIO interrupt problem, which
would fit the "things go horribly wrong for no good reason" category
rather nicely, but it didn't solve it :-( The patch that illustrates
how to switch a function block's interrupt from edge to level is
below.

- Werner

---------------------------------- cut here -----------------------------------

diff --git a/arch/arm/plat-s3c24xx/irq.c b/arch/arm/plat-s3c24xx/irq.c
index f0ed85c..57de24c 100644
--- a/arch/arm/plat-s3c24xx/irq.c
+++ b/arch/arm/plat-s3c24xx/irq.c
@@ -734,6 +734,9 @@ void __init s3c24xx_init_irq(void)
 		switch (irqno) {
 			/* deal with the special IRQs (cascaded) */
 
+		case IRQ_SDI:
+			set_irq_flags(irqno, IRQF_VALID);
+			/* fall through */
 		case IRQ_EINT4t7:
 		case IRQ_EINT8t23:
 		case IRQ_UART0: