ACPI / EC: Don't count a SCI interrupt as a false one

commit a3cd8d2789c2e265e09377f260e7d2ac9cec81bb upstream.

Currently when advance_transaction() is called in EC interrupt handler,
if there is nothing driver can do with the interrupt, it will be taken
as a false one.

But this is not always true, as there may be a SCI EC interrupt fired
during normal read/write operation, which should not be counted as a
false one. This patch fixes the problem.

Signed-off-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
index 14111a4..f553c1c 100644 (file)
--- a/drivers/acpi/ec.c
+++ b/drivers/acpi/ec.c
@@ -205,9 +205,13 @@ static void advance_transaction(struct acpi_ec *ec)
                t->done = true;
        goto unlock;
 err:
-       /* false interrupt, state didn't change */
-       if (in_interrupt())
+       /*
+        * If SCI bit is set, then don't think it's a false IRQ
+        * otherwise will take a not handled IRQ as a false one.
+        */
+       if (in_interrupt() && !(status & ACPI_EC_FLAG_SCI))
                ++t->irq_count;
+
 unlock:
        spin_unlock_irqrestore(&ec->curr_lock, flags);
 }