mfd: ab8500-core: Ignore masked out interrupts

Commit Message

Fabio Baltieri March 21, 2013, 1:49 p.m. UTC

AB8500 asserts LATCH bits for masked out interrupts.  This patch
explicitly masks those out using the cached mask value to prevent
handle_nested_irq() being called for masked IRQ on the same register as
unmasked ones.

Acked-by: Linus Walleij <[e-mail]>
Signed-off-by: Fabio Baltieri <[e-mail]>
---

Hello Samuel, Anton,

this is a small fix for the AB8500 irq generation code.  As this depends
on Lee's patches already merged in the battery tree, this patch is based
on Anton's repository, and should apply cleanly on it and on next.

Samuel, does it make sense to get your ack for this patch and have it
merged with the others in the battery tree?

Thanks,
Fabio

 drivers/mfd/ab8500-core.c | 27 ++++++++++++++-------------
 1 file changed, 14 insertions(+), 13 deletions(-)

Comments

Linus Walleij March 27, 2013, 12:54 p.m. UTC | #1

On Thu, Mar 21, 2013 at 2:49 PM, Fabio Baltieri
<[e-mail]> wrote:

> AB8500 asserts LATCH bits for masked out interrupts.  This patch
> explicitly masks those out using the cached mask value to prevent
> handle_nested_irq() being called for masked IRQ on the same register as
> unmasked ones.
>
> Acked-by: Linus Walleij <[e-mail]>
> Signed-off-by: Fabio Baltieri <[e-mail]>

Sam, ping on this.

Yours,
Linus Walleij

Samuel Ortiz April 9, 2013, 8:20 a.m. UTC | #2

Hi Fabio,

On Thu, Mar 21, 2013 at 02:49:44PM +0100, Fabio Baltieri wrote:
> AB8500 asserts LATCH bits for masked out interrupts.  This patch
> explicitly masks those out using the cached mask value to prevent
> handle_nested_irq() being called for masked IRQ on the same register as
> unmasked ones.
> 
> Acked-by: Linus Walleij <[e-mail]>
> Signed-off-by: Fabio Baltieri <[e-mail]>
> ---
> 
> Hello Samuel, Anton,
> 
> this is a small fix for the AB8500 irq generation code.  As this depends
> on Lee's patches already merged in the battery tree, this patch is based
> on Anton's repository, and should apply cleanly on it and on next.
As Lee prepared a branch for Anton and me to pull from, this patch applied
cleanly on top of mfd-next. So I took it and pushed it now.

Cheers,
Samuel.

Patch

diff --git a/drivers/mfd/ab8500-core.c b/drivers/mfd/ab8500-core.c
index f276352..36751f3 100644
--- a/drivers/mfd/ab8500-core.c
+++ b/drivers/mfd/ab8500-core.c
@@ -458,22 +458,23 @@  static void update_latch_offset(u8 *offset, int i)
 static int ab8500_handle_hierarchical_line(struct ab8500 *ab8500,
 					int latch_offset, u8 latch_val)
 {
-	int int_bit = __ffs(latch_val);
-	int line, i;
+	int int_bit, line, i;
 
-	do {
-		int_bit = __ffs(latch_val);
+	for (i = 0; i < ab8500->mask_size; i++)
+		if (ab8500->irq_reg_offset[i] == latch_offset)
+			break;
 
-		for (i = 0; i < ab8500->mask_size; i++)
-			if (ab8500->irq_reg_offset[i] == latch_offset)
-				break;
+	if (i >= ab8500->mask_size) {
+		dev_err(ab8500->dev, "Register offset 0x%2x not declared\n",
+				latch_offset);
+		return -ENXIO;
+	}
 
-		if (i >= ab8500->mask_size) {
-			dev_err(ab8500->dev, "Register offset 0x%2x not declared\n",
-					latch_offset);
-			return -ENXIO;
-		}
+	/* ignore masked out interrupts */
+	latch_val &= ~ab8500->mask[i];
 
+	while (latch_val) {
+		int_bit = __ffs(latch_val);
 		line = (i << 3) + int_bit;
 		latch_val &= ~(1 << int_bit);
 
@@ -491,7 +492,7 @@  static int ab8500_handle_hierarchical_line(struct ab8500 *ab8500,
 			line += 1;
 
 		handle_nested_irq(ab8500->irq_base + line);
-	} while (latch_val);
+	}
 
 	return 0;
 }