ACPI / ACPICA: Defer enabling of runtime GPEs (v3)

The current ACPI GPEs initialization code has a problem that it
enables some GPEs pointed to by device _PRW methods, generally
intended for signaling wakeup events (system or device wakeup).
These GPEs are then almost immediately disabled by the ACPI namespace
scanning code with the help of acpi_gpe_can_wake(), but it would be
better not to enable them at all until really necessary.

Modify the initialization of GPEs so that the ones that have
associated _Lxx or _Exx methods and are not pointed to by any _PRW
methods will be enabled after the namespace scan is complete.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Len Brown <len.brown@intel.com>
This commit is contained in:
Rafael J. Wysocki
2010-09-16 00:30:43 +02:00
کامیت شده توسط Len Brown
والد 2422084a94
کامیت a210080195
13فایلهای تغییر یافته به همراه94 افزوده شده و 140 حذف شده

مشاهده پرونده

@@ -210,8 +210,7 @@ acpi_status acpi_ev_gpe_initialize(void)
*
* DESCRIPTION: Check for new GPE methods (_Lxx/_Exx) made available as a
* result of a Load() or load_table() operation. If new GPE
* methods have been installed, register the new methods and
* enable and runtime GPEs that are associated with them.
* methods have been installed, register the new methods.
*
******************************************************************************/
@@ -239,7 +238,6 @@ void acpi_ev_update_gpes(acpi_owner_id table_owner_id)
walk_info.owner_id = table_owner_id;
walk_info.execute_by_owner_id = TRUE;
walk_info.count = 0;
walk_info.enable_this_gpe = TRUE;
/* Walk the interrupt level descriptor list */
@@ -301,8 +299,6 @@ void acpi_ev_update_gpes(acpi_owner_id table_owner_id)
*
* If walk_info->execute_by_owner_id is TRUE, we only execute examine GPE methods
* with that owner.
* If walk_info->enable_this_gpe is TRUE, the GPE that is referred to by a GPE
* method is immediately enabled (Used for Load/load_table operators)
*
******************************************************************************/
@@ -315,8 +311,6 @@ acpi_ev_match_gpe_method(acpi_handle obj_handle,
struct acpi_gpe_walk_info *walk_info =
ACPI_CAST_PTR(struct acpi_gpe_walk_info, context);
struct acpi_gpe_event_info *gpe_event_info;
struct acpi_namespace_node *gpe_device;
acpi_status status;
u32 gpe_number;
char name[ACPI_NAME_SIZE + 1];
u8 type;
@@ -421,29 +415,6 @@ acpi_ev_match_gpe_method(acpi_handle obj_handle,
gpe_event_info->flags |= (u8)(type | ACPI_GPE_DISPATCH_METHOD);
gpe_event_info->dispatch.method_node = method_node;
/*
* Enable this GPE if requested. This only happens when during the
* execution of a Load or load_table operator. We have found a new
* GPE method and want to immediately enable the GPE if it is a
* runtime GPE.
*/
if (walk_info->enable_this_gpe) {
walk_info->count++;
gpe_device = walk_info->gpe_device;
if (gpe_device == acpi_gbl_fadt_gpe_device) {
gpe_device = NULL;
}
status = acpi_enable_gpe(gpe_device, gpe_number);
if (ACPI_FAILURE(status)) {
ACPI_EXCEPTION((AE_INFO, status,
"Could not enable GPE 0x%02X",
gpe_number));
}
}
ACPI_DEBUG_PRINT((ACPI_DB_LOAD,
"Registered GPE method %s as GPE number 0x%.2X\n",
name, gpe_number));