lynxpoint: fix enable_pm1() function
The new enable_pm1() function was doing 2 things wrong:
1. It was doing a RMW of the pm1 register. This means we were
keeping around the enables from the OS during S3 resume. This
is bad in the face of the RTC alarm waking us up because it would
cause an infinite stream of SMIs.
2. The register size of PM1_EN is 16-bits. However, the previous
implementation was accessing it as a 32-bit register.
The PM1 enables should only be set to what we expect to handle in the
firmware before the OS changes to ACPI mode.
Change-Id: Ib1d3caf6c84a1670d9456ed159420c6cb64f555e
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: http://review.coreboot.org/2978
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
diff --git a/src/southbridge/intel/lynxpoint/pch.h b/src/southbridge/intel/lynxpoint/pch.h
index 3535b98..659ddb6 100644
--- a/src/southbridge/intel/lynxpoint/pch.h
+++ b/src/southbridge/intel/lynxpoint/pch.h
@@ -139,7 +139,7 @@
void disable_pm1_control(u32 mask);
/* PM1 */
u16 clear_pm1_status(void);
-void enable_pm1(u32 mask);
+void enable_pm1(u16 events);
u32 clear_smi_status(void);
/* SMI */
void enable_smi(u32 mask);