summaryrefslogtreecommitdiffstats
path: root/drivers/acpi
diff options
context:
space:
mode:
authorLen Brown <len.brown@intel.com>2005-07-30 05:31:17 +0200
committerLen Brown <len.brown@intel.com>2005-07-30 05:31:17 +0200
commitd6ac1a7910d22626bc77e73db091e00b810715f4 (patch)
treec8f235b7287056a03bc2bc032902469a331b5e56 /drivers/acpi
parent[PATCH] More qla2xxx configuration fixes (diff)
parent[ACPI] suspend/resume ACPI PCI Interrupt Links (diff)
downloadlinux-d6ac1a7910d22626bc77e73db091e00b810715f4.tar.xz
linux-d6ac1a7910d22626bc77e73db091e00b810715f4.zip
/home/lenb/src/to-linus branch 'acpi-2.6.12'
Diffstat (limited to 'drivers/acpi')
-rw-r--r--drivers/acpi/ec.c893
-rw-r--r--drivers/acpi/pci_irq.c85
-rw-r--r--drivers/acpi/pci_link.c103
-rw-r--r--drivers/acpi/processor_idle.c31
4 files changed, 885 insertions, 227 deletions
diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c
index fca4140a50a9..2dadb7f63269 100644
--- a/drivers/acpi/ec.c
+++ b/drivers/acpi/ec.c
@@ -59,76 +59,185 @@ ACPI_MODULE_NAME ("acpi_ec")
#define ACPI_EC_DELAY 50 /* Wait 50ms max. during EC ops */
#define ACPI_EC_UDELAY_GLK 1000 /* Wait 1ms max. to get global lock */
+#define ACPI_EC_UDELAY 100 /* Poll @ 100us increments */
+#define ACPI_EC_UDELAY_COUNT 1000 /* Wait 10ms max. during EC ops */
+
#define ACPI_EC_COMMAND_READ 0x80
#define ACPI_EC_COMMAND_WRITE 0x81
#define ACPI_EC_BURST_ENABLE 0x82
#define ACPI_EC_BURST_DISABLE 0x83
#define ACPI_EC_COMMAND_QUERY 0x84
-static int acpi_ec_add (struct acpi_device *device);
+#define EC_POLLING 0xFF
+#define EC_BURST 0x00
+
+
static int acpi_ec_remove (struct acpi_device *device, int type);
static int acpi_ec_start (struct acpi_device *device);
static int acpi_ec_stop (struct acpi_device *device, int type);
+static int acpi_ec_burst_add ( struct acpi_device *device);
static struct acpi_driver acpi_ec_driver = {
.name = ACPI_EC_DRIVER_NAME,
.class = ACPI_EC_CLASS,
.ids = ACPI_EC_HID,
.ops = {
- .add = acpi_ec_add,
+ .add = acpi_ec_burst_add,
.remove = acpi_ec_remove,
.start = acpi_ec_start,
.stop = acpi_ec_stop,
},
};
-
-struct acpi_ec {
- acpi_handle handle;
- unsigned long uid;
- unsigned long gpe_bit;
- struct acpi_generic_address status_addr;
- struct acpi_generic_address command_addr;
- struct acpi_generic_address data_addr;
- unsigned long global_lock;
- unsigned int expect_event;
- atomic_t leaving_burst; /* 0 : No, 1 : Yes, 2: abort*/
- atomic_t pending_gpe;
- struct semaphore sem;
- wait_queue_head_t wait;
+union acpi_ec {
+ struct {
+ u32 mode;
+ acpi_handle handle;
+ unsigned long uid;
+ unsigned long gpe_bit;
+ struct acpi_generic_address status_addr;
+ struct acpi_generic_address command_addr;
+ struct acpi_generic_address data_addr;
+ unsigned long global_lock;
+ } common;
+
+ struct {
+ u32 mode;
+ acpi_handle handle;
+ unsigned long uid;
+ unsigned long gpe_bit;
+ struct acpi_generic_address status_addr;
+ struct acpi_generic_address command_addr;
+ struct acpi_generic_address data_addr;
+ unsigned long global_lock;
+ unsigned int expect_event;
+ atomic_t leaving_burst; /* 0 : No, 1 : Yes, 2: abort*/
+ atomic_t pending_gpe;
+ struct semaphore sem;
+ wait_queue_head_t wait;
+ }burst;
+
+ struct {
+ u32 mode;
+ acpi_handle handle;
+ unsigned long uid;
+ unsigned long gpe_bit;
+ struct acpi_generic_address status_addr;
+ struct acpi_generic_address command_addr;
+ struct acpi_generic_address data_addr;
+ unsigned long global_lock;
+ spinlock_t lock;
+ }polling;
};
+static int acpi_ec_polling_wait ( union acpi_ec *ec, u8 event);
+static int acpi_ec_burst_wait(union acpi_ec *ec, unsigned int event);
+static int acpi_ec_polling_read ( union acpi_ec *ec, u8 address, u32 *data);
+static int acpi_ec_burst_read( union acpi_ec *ec, u8 address, u32 *data);
+static int acpi_ec_polling_write ( union acpi_ec *ec, u8 address, u8 data);
+static int acpi_ec_burst_write ( union acpi_ec *ec, u8 address, u8 data);
+static int acpi_ec_polling_query ( union acpi_ec *ec, u32 *data);
+static int acpi_ec_burst_query ( union acpi_ec *ec, u32 *data);
+static void acpi_ec_gpe_polling_query ( void *ec_cxt);
+static void acpi_ec_gpe_burst_query ( void *ec_cxt);
+static u32 acpi_ec_gpe_polling_handler ( void *data);
+static u32 acpi_ec_gpe_burst_handler ( void *data);
+static acpi_status __init
+acpi_fake_ecdt_polling_callback (
+ acpi_handle handle,
+ u32 Level,
+ void *context,
+ void **retval);
+
+static acpi_status __init
+acpi_fake_ecdt_burst_callback (
+ acpi_handle handle,
+ u32 Level,
+ void *context,
+ void **retval);
+
+static int __init
+acpi_ec_polling_get_real_ecdt(void);
+static int __init
+acpi_ec_burst_get_real_ecdt(void);
/* If we find an EC via the ECDT, we need to keep a ptr to its context */
-static struct acpi_ec *ec_ecdt;
+static union acpi_ec *ec_ecdt;
/* External interfaces use first EC only, so remember */
static struct acpi_device *first_ec;
+static int acpi_ec_polling_mode;
/* --------------------------------------------------------------------------
Transaction Management
-------------------------------------------------------------------------- */
-static inline u32 acpi_ec_read_status(struct acpi_ec *ec)
+static inline u32 acpi_ec_read_status(union acpi_ec *ec)
{
u32 status = 0;
- acpi_hw_low_level_read(8, &status, &ec->status_addr);
+ acpi_hw_low_level_read(8, &status, &ec->common.status_addr);
return status;
}
-static int acpi_ec_wait(struct acpi_ec *ec, unsigned int event)
+static int
+acpi_ec_wait (
+ union acpi_ec *ec,
+ u8 event)
+{
+ if (acpi_ec_polling_mode)
+ return acpi_ec_polling_wait (ec, event);
+ else
+ return acpi_ec_burst_wait (ec, event);
+}
+
+static int
+acpi_ec_polling_wait (
+ union acpi_ec *ec,
+ u8 event)
+{
+ u32 acpi_ec_status = 0;
+ u32 i = ACPI_EC_UDELAY_COUNT;
+
+ if (!ec)
+ return -EINVAL;
+
+ /* Poll the EC status register waiting for the event to occur. */
+ switch (event) {
+ case ACPI_EC_EVENT_OBF:
+ do {
+ acpi_hw_low_level_read(8, &acpi_ec_status, &ec->common.status_addr);
+ if (acpi_ec_status & ACPI_EC_FLAG_OBF)
+ return 0;
+ udelay(ACPI_EC_UDELAY);
+ } while (--i>0);
+ break;
+ case ACPI_EC_EVENT_IBE:
+ do {
+ acpi_hw_low_level_read(8, &acpi_ec_status, &ec->common.status_addr);
+ if (!(acpi_ec_status & ACPI_EC_FLAG_IBF))
+ return 0;
+ udelay(ACPI_EC_UDELAY);
+ } while (--i>0);
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return -ETIME;
+}
+static int acpi_ec_burst_wait(union acpi_ec *ec, unsigned int event)
{
int result = 0;
ACPI_FUNCTION_TRACE("acpi_ec_wait");
- ec->expect_event = event;
+ ec->burst.expect_event = event;
smp_mb();
- result = wait_event_interruptible_timeout(ec->wait,
- !ec->expect_event,
+ result = wait_event_interruptible_timeout(ec->burst.wait,
+ !ec->burst.expect_event,
msecs_to_jiffies(ACPI_EC_DELAY));
- ec->expect_event = 0;
+ ec->burst.expect_event = 0;
smp_mb();
if (result < 0){
@@ -160,7 +269,7 @@ static int acpi_ec_wait(struct acpi_ec *ec, unsigned int event)
static int
acpi_ec_enter_burst_mode (
- struct acpi_ec *ec)
+ union acpi_ec *ec)
{
u32 tmp = 0;
int status = 0;
@@ -170,43 +279,43 @@ acpi_ec_enter_burst_mode (
status = acpi_ec_read_status(ec);
if (status != -EINVAL &&
!(status & ACPI_EC_FLAG_BURST)){
- acpi_hw_low_level_write(8, ACPI_EC_BURST_ENABLE, &ec->command_addr);
+ acpi_hw_low_level_write(8, ACPI_EC_BURST_ENABLE, &ec->common.command_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_OBF);
if (status){
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
return_VALUE(-EINVAL);
}
- acpi_hw_low_level_read(8, &tmp, &ec->data_addr);
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_hw_low_level_read(8, &tmp, &ec->common.data_addr);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
if(tmp != 0x90 ) {/* Burst ACK byte*/
return_VALUE(-EINVAL);
}
}
- atomic_set(&ec->leaving_burst , 0);
+ atomic_set(&ec->burst.leaving_burst , 0);
return_VALUE(0);
}
static int
acpi_ec_leave_burst_mode (
- struct acpi_ec *ec)
+ union acpi_ec *ec)
{
int status =0;
ACPI_FUNCTION_TRACE("acpi_ec_leave_burst_mode");
- atomic_set(&ec->leaving_burst , 1);
+ atomic_set(&ec->burst.leaving_burst , 1);
status = acpi_ec_read_status(ec);
if (status != -EINVAL &&
(status & ACPI_EC_FLAG_BURST)){
- acpi_hw_low_level_write(8, ACPI_EC_BURST_DISABLE, &ec->command_addr);
+ acpi_hw_low_level_write(8, ACPI_EC_BURST_DISABLE, &ec->common.command_addr);
status = acpi_ec_wait(ec, ACPI_EC_FLAG_IBF);
if (status){
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
ACPI_DEBUG_PRINT((ACPI_DB_ERROR,"------->wait fail\n"));
return_VALUE(-EINVAL);
}
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
status = acpi_ec_read_status(ec);
}
@@ -215,7 +324,131 @@ acpi_ec_leave_burst_mode (
static int
acpi_ec_read (
- struct acpi_ec *ec,
+ union acpi_ec *ec,
+ u8 address,
+ u32 *data)
+{
+ if (acpi_ec_polling_mode)
+ return acpi_ec_polling_read(ec, address, data);
+ else
+ return acpi_ec_burst_read(ec, address, data);
+}
+static int
+acpi_ec_write (
+ union acpi_ec *ec,
+ u8 address,
+ u8 data)
+{
+ if (acpi_ec_polling_mode)
+ return acpi_ec_polling_write(ec, address, data);
+ else
+ return acpi_ec_burst_write(ec, address, data);
+}
+static int
+acpi_ec_polling_read (
+ union acpi_ec *ec,
+ u8 address,
+ u32 *data)
+{
+ acpi_status status = AE_OK;
+ int result = 0;
+ unsigned long flags = 0;
+ u32 glk = 0;
+
+ ACPI_FUNCTION_TRACE("acpi_ec_read");
+
+ if (!ec || !data)
+ return_VALUE(-EINVAL);
+
+ *data = 0;
+
+ if (ec->common.global_lock) {
+ status = acpi_acquire_global_lock(ACPI_EC_UDELAY_GLK, &glk);
+ if (ACPI_FAILURE(status))
+ return_VALUE(-ENODEV);
+ }
+
+ spin_lock_irqsave(&ec->polling.lock, flags);
+
+ acpi_hw_low_level_write(8, ACPI_EC_COMMAND_READ, &ec->common.command_addr);
+ result = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
+ if (result)
+ goto end;
+
+ acpi_hw_low_level_write(8, address, &ec->common.data_addr);
+ result = acpi_ec_wait(ec, ACPI_EC_EVENT_OBF);
+ if (result)
+ goto end;
+
+ acpi_hw_low_level_read(8, data, &ec->common.data_addr);
+
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Read [%02x] from address [%02x]\n",
+ *data, address));
+
+end:
+ spin_unlock_irqrestore(&ec->polling.lock, flags);
+
+ if (ec->common.global_lock)
+ acpi_release_global_lock(glk);
+
+ return_VALUE(result);
+}
+
+
+static int
+acpi_ec_polling_write (
+ union acpi_ec *ec,
+ u8 address,
+ u8 data)
+{
+ int result = 0;
+ acpi_status status = AE_OK;
+ unsigned long flags = 0;
+ u32 glk = 0;
+
+ ACPI_FUNCTION_TRACE("acpi_ec_write");
+
+ if (!ec)
+ return_VALUE(-EINVAL);
+
+ if (ec->common.global_lock) {
+ status = acpi_acquire_global_lock(ACPI_EC_UDELAY_GLK, &glk);
+ if (ACPI_FAILURE(status))
+ return_VALUE(-ENODEV);
+ }
+
+ spin_lock_irqsave(&ec->polling.lock, flags);
+
+ acpi_hw_low_level_write(8, ACPI_EC_COMMAND_WRITE, &ec->common.command_addr);
+ result = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
+ if (result)
+ goto end;
+
+ acpi_hw_low_level_write(8, address, &ec->common.data_addr);
+ result = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
+ if (result)
+ goto end;
+
+ acpi_hw_low_level_write(8, data, &ec->common.data_addr);
+ result = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
+ if (result)
+ goto end;
+
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Wrote [%02x] to address [%02x]\n",
+ data, address));
+
+end:
+ spin_unlock_irqrestore(&ec->polling.lock, flags);
+
+ if (ec->common.global_lock)
+ acpi_release_global_lock(glk);
+
+ return_VALUE(result);
+}
+
+static int
+acpi_ec_burst_read (
+ union acpi_ec *ec,
u8 address,
u32 *data)
{
@@ -230,51 +463,51 @@ acpi_ec_read (
retry:
*data = 0;
- if (ec->global_lock) {
+ if (ec->common.global_lock) {
status = acpi_acquire_global_lock(ACPI_EC_UDELAY_GLK, &glk);
if (ACPI_FAILURE(status))
return_VALUE(-ENODEV);
}
WARN_ON(in_interrupt());
- down(&ec->sem);
+ down(&ec->burst.sem);
if(acpi_ec_enter_burst_mode(ec))
goto end;
- acpi_hw_low_level_write(8, ACPI_EC_COMMAND_READ, &ec->command_addr);
+ acpi_hw_low_level_write(8, ACPI_EC_COMMAND_READ, &ec->common.command_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
if (status) {
goto end;
}
- acpi_hw_low_level_write(8, address, &ec->data_addr);
+ acpi_hw_low_level_write(8, address, &ec->common.data_addr);
status= acpi_ec_wait(ec, ACPI_EC_EVENT_OBF);
if (status){
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
goto end;
}
- acpi_hw_low_level_read(8, data, &ec->data_addr);
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_hw_low_level_read(8, data, &ec->common.data_addr);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Read [%02x] from address [%02x]\n",
*data, address));
end:
acpi_ec_leave_burst_mode(ec);
- up(&ec->sem);
+ up(&ec->burst.sem);
- if (ec->global_lock)
+ if (ec->common.global_lock)
acpi_release_global_lock(glk);
- if(atomic_read(&ec->leaving_burst) == 2){
+ if(atomic_read(&ec->burst.leaving_burst) == 2){
ACPI_DEBUG_PRINT((ACPI_DB_INFO,"aborted, retry ...\n"));
- while(atomic_read(&ec->pending_gpe)){
+ while(atomic_read(&ec->burst.pending_gpe)){
msleep(1);
}
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
goto retry;
}
@@ -283,8 +516,8 @@ end:
static int
-acpi_ec_write (
- struct acpi_ec *ec,
+acpi_ec_burst_write (
+ union acpi_ec *ec,
u8 address,
u8 data)
{
@@ -297,14 +530,14 @@ acpi_ec_write (
if (!ec)
return_VALUE(-EINVAL);
retry:
- if (ec->global_lock) {
+ if (ec->common.global_lock) {
status = acpi_acquire_global_lock(ACPI_EC_UDELAY_GLK, &glk);
if (ACPI_FAILURE(status))
return_VALUE(-ENODEV);
}
WARN_ON(in_interrupt());
- down(&ec->sem);
+ down(&ec->burst.sem);
if(acpi_ec_enter_burst_mode(ec))
goto end;
@@ -312,33 +545,33 @@ retry:
status = acpi_ec_read_status(ec);
if (status != -EINVAL &&
!(status & ACPI_EC_FLAG_BURST)){
- acpi_hw_low_level_write(8, ACPI_EC_BURST_ENABLE, &ec->command_addr);
+ acpi_hw_low_level_write(8, ACPI_EC_BURST_ENABLE, &ec->common.command_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_OBF);
if (status)
goto end;
- acpi_hw_low_level_read(8, &tmp, &ec->data_addr);
+ acpi_hw_low_level_read(8, &tmp, &ec->common.data_addr);
if(tmp != 0x90 ) /* Burst ACK byte*/
goto end;
}
/*Now we are in burst mode*/
- acpi_hw_low_level_write(8, ACPI_EC_COMMAND_WRITE, &ec->command_addr);
+ acpi_hw_low_level_write(8, ACPI_EC_COMMAND_WRITE, &ec->common.command_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
if (status){
goto end;
}
- acpi_hw_low_level_write(8, address, &ec->data_addr);
+ acpi_hw_low_level_write(8, address, &ec->common.data_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
if (status){
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
goto end;
}
- acpi_hw_low_level_write(8, data, &ec->data_addr);
+ acpi_hw_low_level_write(8, data, &ec->common.data_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_IBE);
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
if (status)
goto end;
@@ -347,17 +580,17 @@ retry:
end:
acpi_ec_leave_burst_mode(ec);
- up(&ec->sem);
+ up(&ec->burst.sem);
- if (ec->global_lock)
+ if (ec->common.global_lock)
acpi_release_global_lock(glk);
- if(atomic_read(&ec->leaving_burst) == 2){
+ if(atomic_read(&ec->burst.leaving_burst) == 2){
ACPI_DEBUG_PRINT((ACPI_DB_INFO,"aborted, retry ...\n"));
- while(atomic_read(&ec->pending_gpe)){
+ while(atomic_read(&ec->burst.pending_gpe)){
msleep(1);
}
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
goto retry;
}
@@ -370,7 +603,7 @@ end:
int
ec_read(u8 addr, u8 *val)
{
- struct acpi_ec *ec;
+ union acpi_ec *ec;
int err;
u32 temp_data;
@@ -393,7 +626,7 @@ EXPORT_SYMBOL(ec_read);
int
ec_write(u8 addr, u8 val)
{
- struct acpi_ec *ec;
+ union acpi_ec *ec;
int err;
if (!first_ec)
@@ -407,10 +640,66 @@ ec_write(u8 addr, u8 val)
}
EXPORT_SYMBOL(ec_write);
-
static int
acpi_ec_query (
- struct acpi_ec *ec,
+ union acpi_ec *ec,
+ u32 *data)
+{
+ if (acpi_ec_polling_mode)
+ return acpi_ec_polling_query(ec, data);
+ else
+ return acpi_ec_burst_query(ec, data);
+}
+static int
+acpi_ec_polling_query (
+ union acpi_ec *ec,
+ u32 *data)
+{
+ int result = 0;
+ acpi_status status = AE_OK;
+ unsigned long flags = 0;
+ u32 glk = 0;
+
+ ACPI_FUNCTION_TRACE("acpi_ec_query");
+
+ if (!ec || !data)
+ return_VALUE(-EINVAL);
+
+ *data = 0;
+
+ if (ec->common.global_lock) {
+ status = acpi_acquire_global_lock(ACPI_EC_UDELAY_GLK, &glk);
+ if (ACPI_FAILURE(status))
+ return_VALUE(-ENODEV);
+ }
+
+ /*
+ * Query the EC to find out which _Qxx method we need to evaluate.
+ * Note that successful completion of the query causes the ACPI_EC_SCI
+ * bit to be cleared (and thus clearing the interrupt source).
+ */
+ spin_lock_irqsave(&ec->polling.lock, flags);
+
+ acpi_hw_low_level_write(8, ACPI_EC_COMMAND_QUERY, &ec->common.command_addr);
+ result = acpi_ec_wait(ec, ACPI_EC_EVENT_OBF);
+ if (result)
+ goto end;
+
+ acpi_hw_low_level_read(8, data, &ec->common.data_addr);
+ if (!*data)
+ result = -ENODATA;
+
+end:
+ spin_unlock_irqrestore(&ec->polling.lock, flags);
+
+ if (ec->common.global_lock)
+ acpi_release_global_lock(glk);
+
+ return_VALUE(result);
+}
+static int
+acpi_ec_burst_query (
+ union acpi_ec *ec,
u32 *data)
{
int status = 0;
@@ -422,13 +711,13 @@ acpi_ec_query (
return_VALUE(-EINVAL);
*data = 0;
- if (ec->global_lock) {
+ if (ec->common.global_lock) {
status = acpi_acquire_global_lock(ACPI_EC_UDELAY_GLK, &glk);
if (ACPI_FAILURE(status))
return_VALUE(-ENODEV);
}
- down(&ec->sem);
+ down(&ec->burst.sem);
if(acpi_ec_enter_burst_mode(ec))
goto end;
/*
@@ -436,28 +725,28 @@ acpi_ec_query (
* Note that successful completion of the query causes the ACPI_EC_SCI
* bit to be cleared (and thus clearing the interrupt source).
*/
- acpi_hw_low_level_write(8, ACPI_EC_COMMAND_QUERY, &ec->command_addr);
+ acpi_hw_low_level_write(8, ACPI_EC_COMMAND_QUERY, &ec->common.command_addr);
status = acpi_ec_wait(ec, ACPI_EC_EVENT_OBF);
if (status){
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
goto end;
}
- acpi_hw_low_level_read(8, data, &ec->data_addr);
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_hw_low_level_read(8, data, &ec->common.data_addr);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
if (!*data)
status = -ENODATA;
end:
acpi_ec_leave_burst_mode(ec);
- up(&ec->sem);
+ up(&ec->burst.sem);
- if (ec->global_lock)
+ if (ec->common.global_lock)
acpi_release_global_lock(glk);
- if(atomic_read(&ec->leaving_burst) == 2){
+ if(atomic_read(&ec->burst.leaving_burst) == 2){
ACPI_DEBUG_PRINT((ACPI_DB_INFO,"aborted, retry ...\n"));
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
status = -ENODATA;
}
return_VALUE(status);
@@ -468,7 +757,7 @@ end:
Event Management
-------------------------------------------------------------------------- */
-struct acpi_ec_query_data {
+union acpi_ec_query_data {
acpi_handle handle;
u8 data;
};
@@ -477,7 +766,59 @@ static void
acpi_ec_gpe_query (
void *ec_cxt)
{
- struct acpi_ec *ec = (struct acpi_ec *) ec_cxt;
+ if (acpi_ec_polling_mode)
+ acpi_ec_gpe_polling_query(ec_cxt);
+ else
+ acpi_ec_gpe_burst_query(ec_cxt);
+}
+
+static void
+acpi_ec_gpe_polling_query (
+ void *ec_cxt)
+{
+ union acpi_ec *ec = (union acpi_ec *) ec_cxt;
+ u32 value = 0;
+ unsigned long flags = 0;
+ static char object_name[5] = {'_','Q','0','0','\0'};
+ const char hex[] = {'0','1','2','3','4','5','6','7',
+ '8','9','A','B','C','D','E','F'};
+
+ ACPI_FUNCTION_TRACE("acpi_ec_gpe_query");
+
+ if (!ec_cxt)
+ goto end;
+
+ spin_lock_irqsave(&ec->polling.lock, flags);
+ acpi_hw_low_level_read(8, &value, &ec->common.command_addr);
+ spin_unlock_irqrestore(&ec->polling.lock, flags);
+
+ /* TBD: Implement asynch events!
+ * NOTE: All we care about are EC-SCI's. Other EC events are
+ * handled via polling (yuck!). This is because some systems
+ * treat EC-SCIs as level (versus EDGE!) triggered, preventing
+ * a purely interrupt-driven approach (grumble, grumble).
+ */
+ if (!(value & ACPI_EC_FLAG_SCI))
+ goto end;
+
+ if (acpi_ec_query(ec, &value))
+ goto end;
+
+ object_name[2] = hex[((value >> 4) & 0x0F)];
+ object_name[3] = hex[(value & 0x0F)];
+
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Evaluating %s\n", object_name));
+
+ acpi_evaluate_object(ec->common.handle, object_name, NULL, NULL);
+
+end:
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
+}
+static void
+acpi_ec_gpe_burst_query (
+ void *ec_cxt)
+{
+ union acpi_ec *ec = (union acpi_ec *) ec_cxt;
u32 value;
int result = -ENODATA;
static char object_name[5] = {'_','Q','0','0','\0'};
@@ -497,9 +838,9 @@ acpi_ec_gpe_query (
ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Evaluating %s\n", object_name));
- acpi_evaluate_object(ec->handle, object_name, NULL, NULL);
+ acpi_evaluate_object(ec->common.handle, object_name, NULL, NULL);
end:
- atomic_dec(&ec->pending_gpe);
+ atomic_dec(&ec->burst.pending_gpe);
return;
}
@@ -507,48 +848,77 @@ static u32
acpi_ec_gpe_handler (
void *data)
{
+ if (acpi_ec_polling_mode)
+ return acpi_ec_gpe_polling_handler(data);
+ else
+ return acpi_ec_gpe_burst_handler(data);
+}
+static u32
+acpi_ec_gpe_polling_handler (
+ void *data)
+{
+ acpi_status status = AE_OK;
+ union acpi_ec *ec = (union acpi_ec *) data;
+
+ if (!ec)
+ return ACPI_INTERRUPT_NOT_HANDLED;
+
+ acpi_disable_gpe(NULL, ec->common.gpe_bit, ACPI_ISR);
+
+ status = acpi_os_queue_for_execution(OSD_PRIORITY_GPE,
+ acpi_ec_gpe_query, ec);
+
+ if (status == AE_OK)
+ return ACPI_INTERRUPT_HANDLED;
+ else
+ return ACPI_INTERRUPT_NOT_HANDLED;
+}
+static u32
+acpi_ec_gpe_burst_handler (
+ void *data)
+{
acpi_status status = AE_OK;
u32 value;
- struct acpi_ec *ec = (struct acpi_ec *) data;
+ union acpi_ec *ec = (union acpi_ec *) data;
if (!ec)
return ACPI_INTERRUPT_NOT_HANDLED;
- acpi_disable_gpe(NULL, ec->gpe_bit, ACPI_ISR);
+ acpi_disable_gpe(NULL, ec->common.gpe_bit, ACPI_ISR);
value = acpi_ec_read_status(ec);
if((value & ACPI_EC_FLAG_IBF) &&
!(value & ACPI_EC_FLAG_BURST) &&
- (atomic_read(&ec->leaving_burst) == 0)) {
+ (atomic_read(&ec->burst.leaving_burst) == 0)) {
/*
* the embedded controller disables
* burst mode for any reason other
* than the burst disable command
* to process critical event.
*/
- atomic_set(&ec->leaving_burst , 2); /* block current pending transaction
+ atomic_set(&ec->burst.leaving_burst , 2); /* block current pending transaction
and retry */
- wake_up(&ec->wait);
+ wake_up(&ec->burst.wait);
}else {
- if ((ec->expect_event == ACPI_EC_EVENT_OBF &&
+ if ((ec->burst.expect_event == ACPI_EC_EVENT_OBF &&
(value & ACPI_EC_FLAG_OBF)) ||
- (ec->expect_event == ACPI_EC_EVENT_IBE &&
+ (ec->burst.expect_event == ACPI_EC_EVENT_IBE &&
!(value & ACPI_EC_FLAG_IBF))) {
- ec->expect_event = 0;
- wake_up(&ec->wait);
+ ec->burst.expect_event = 0;
+ wake_up(&ec->burst.wait);
return ACPI_INTERRUPT_HANDLED;
}
}
if (value & ACPI_EC_FLAG_SCI){
- atomic_add(1, &ec->pending_gpe) ;
+ atomic_add(1, &ec->burst.pending_gpe) ;
status = acpi_os_queue_for_execution(OSD_PRIORITY_GPE,
acpi_ec_gpe_query, ec);
return status == AE_OK ?
ACPI_INTERRUPT_HANDLED : ACPI_INTERRUPT_NOT_HANDLED;
}
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_ISR);
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_ISR);
return status == AE_OK ?
ACPI_INTERRUPT_HANDLED : ACPI_INTERRUPT_NOT_HANDLED;
}
@@ -585,7 +955,7 @@ acpi_ec_space_handler (
void *region_context)
{
int result = 0;
- struct acpi_ec *ec = NULL;
+ union acpi_ec *ec = NULL;
u64 temp = *value;
acpi_integer f_v = 0;
int i = 0;
@@ -600,7 +970,7 @@ acpi_ec_space_handler (
return_VALUE(AE_BAD_PARAMETER);
}
- ec = (struct acpi_ec *) handler_context;
+ ec = (union acpi_ec *) handler_context;
next_byte:
switch (function) {
@@ -661,7 +1031,7 @@ static struct proc_dir_entry *acpi_ec_dir;
static int
acpi_ec_read_info (struct seq_file *seq, void *offset)
{
- struct acpi_ec *ec = (struct acpi_ec *) seq->private;
+ union acpi_ec *ec = (union acpi_ec *) seq->private;
ACPI_FUNCTION_TRACE("acpi_ec_read_info");
@@ -669,12 +1039,12 @@ acpi_ec_read_info (struct seq_file *seq, void *offset)
goto end;
seq_printf(seq, "gpe bit: 0x%02x\n",
- (u32) ec->gpe_bit);
+ (u32) ec->common.gpe_bit);
seq_printf(seq, "ports: 0x%02x, 0x%02x\n",
- (u32) ec->status_addr.address, (u32) ec->data_addr.address);
+ (u32) ec->common.status_addr.address, (u32) ec->common.data_addr.address);
seq_printf(seq, "use global lock: %s\n",
- ec->global_lock?"yes":"no");
- acpi_enable_gpe(NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ ec->common.global_lock?"yes":"no");
+ acpi_enable_gpe(NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
end:
return_VALUE(0);
@@ -697,7 +1067,7 @@ static int
acpi_ec_add_fs (
struct acpi_device *device)
{
- struct proc_dir_entry *entry;
+ struct proc_dir_entry *entry = NULL;
ACPI_FUNCTION_TRACE("acpi_ec_add_fs");
@@ -744,13 +1114,82 @@ acpi_ec_remove_fs (
Driver Interface
-------------------------------------------------------------------------- */
+
static int
-acpi_ec_add (
+acpi_ec_polling_add (
struct acpi_device *device)
{
- int result;
- acpi_status status;
- struct acpi_ec *ec;
+ int result = 0;
+ acpi_status status = AE_OK;
+ union acpi_ec *ec = NULL;
+ unsigned long uid;
+
+ ACPI_FUNCTION_TRACE("acpi_ec_add");
+
+ if (!device)
+ return_VALUE(-EINVAL);
+
+ ec = kmalloc(sizeof(union acpi_ec), GFP_KERNEL);
+ if (!ec)
+ return_VALUE(-ENOMEM);
+ memset(ec, 0, sizeof(union acpi_ec));
+
+ ec->common.handle = device->handle;
+ ec->common.uid = -1;
+ spin_lock_init(&ec->polling.lock);
+ strcpy(acpi_device_name(device), ACPI_EC_DEVICE_NAME);
+ strcpy(acpi_device_class(device), ACPI_EC_CLASS);
+ acpi_driver_data(device) = ec;
+
+ /* Use the global lock for all EC transactions? */
+ acpi_evaluate_integer(ec->common.handle, "_GLK", NULL, &ec->common.global_lock);
+
+ /* If our UID matches the UID for the ECDT-enumerated EC,
+ we now have the *real* EC info, so kill the makeshift one.*/
+ acpi_evaluate_integer(ec->common.handle, "_UID", NULL, &uid);
+ if (ec_ecdt && ec_ecdt->common.uid == uid) {
+ acpi_remove_address_space_handler(ACPI_ROOT_OBJECT,
+ ACPI_ADR_SPACE_EC, &acpi_ec_space_handler);
+
+ acpi_remove_gpe_handler(NULL, ec_ecdt->common.gpe_bit, &acpi_ec_gpe_handler);
+
+ kfree(ec_ecdt);
+ }
+
+ /* Get GPE bit assignment (EC events). */
+ /* TODO: Add support for _GPE returning a package */
+ status = acpi_evaluate_integer(ec->common.handle, "_GPE", NULL, &ec->common.gpe_bit);
+ if (ACPI_FAILURE(status)) {
+ ACPI_DEBUG_PRINT((ACPI_DB_ERROR,
+ "Error obtaining GPE bit assignment\n"));
+ result = -ENODEV;
+ goto end;
+ }
+
+ result = acpi_ec_add_fs(device);
+ if (result)
+ goto end;
+
+ printk(KERN_INFO PREFIX "%s [%s] (gpe %d)\n",
+ acpi_device_name(device), acpi_device_bid(device),
+ (u32) ec->common.gpe_bit);
+
+ if (!first_ec)
+ first_ec = device;
+
+end:
+ if (result)
+ kfree(ec);
+
+ return_VALUE(result);
+}
+static int
+acpi_ec_burst_add (
+ struct acpi_device *device)
+{
+ int result = 0;
+ acpi_status status = AE_OK;
+ union acpi_ec *ec = NULL;
unsigned long uid;
ACPI_FUNCTION_TRACE("acpi_ec_add");
@@ -758,39 +1197,39 @@ acpi_ec_add (
if (!device)
return_VALUE(-EINVAL);
- ec = kmalloc(sizeof(struct acpi_ec), GFP_KERNEL);
+ ec = kmalloc(sizeof(union acpi_ec), GFP_KERNEL);
if (!ec)
return_VALUE(-ENOMEM);
- memset(ec, 0, sizeof(struct acpi_ec));
-
- ec->handle = device->handle;
- ec->uid = -1;
- atomic_set(&ec->pending_gpe, 0);
- atomic_set(&ec->leaving_burst , 1);
- init_MUTEX(&ec->sem);
- init_waitqueue_head(&ec->wait);
+ memset(ec, 0, sizeof(union acpi_ec));
+
+ ec->common.handle = device->handle;
+ ec->common.uid = -1;
+ atomic_set(&ec->burst.pending_gpe, 0);
+ atomic_set(&ec->burst.leaving_burst , 1);
+ init_MUTEX(&ec->burst.sem);
+ init_waitqueue_head(&ec->burst.wait);
strcpy(acpi_device_name(device), ACPI_EC_DEVICE_NAME);
strcpy(acpi_device_class(device), ACPI_EC_CLASS);
acpi_driver_data(device) = ec;
/* Use the global lock for all EC transactions? */
- acpi_evaluate_integer(ec->handle, "_GLK", NULL, &ec->global_lock);
+ acpi_evaluate_integer(ec->common.handle, "_GLK", NULL, &ec->common.global_lock);
/* If our UID matches the UID for the ECDT-enumerated EC,
we now have the *real* EC info, so kill the makeshift one.*/
- acpi_evaluate_integer(ec->handle, "_UID", NULL, &uid);
- if (ec_ecdt && ec_ecdt->uid == uid) {
+ acpi_evaluate_integer(ec->common.handle, "_UID", NULL, &uid);
+ if (ec_ecdt && ec_ecdt->common.uid == uid) {
acpi_remove_address_space_handler(ACPI_ROOT_OBJECT,
ACPI_ADR_SPACE_EC, &acpi_ec_space_handler);
- acpi_remove_gpe_handler(NULL, ec_ecdt->gpe_bit, &acpi_ec_gpe_handler);
+ acpi_remove_gpe_handler(NULL, ec_ecdt->common.gpe_bit, &acpi_ec_gpe_handler);
kfree(ec_ecdt);
}
/* Get GPE bit assignment (EC events). */
/* TODO: Add support for _GPE returning a package */
- status = acpi_evaluate_integer(ec->handle, "_GPE", NULL, &ec->gpe_bit);
+ status = acpi_evaluate_integer(ec->common.handle, "_GPE", NULL, &ec->common.gpe_bit);
if (ACPI_FAILURE(status)) {
ACPI_DEBUG_PRINT((ACPI_DB_ERROR,
"Error obtaining GPE bit assignment\n"));
@@ -804,7 +1243,7 @@ acpi_ec_add (
printk(KERN_INFO PREFIX "%s [%s] (gpe %d)\n",
acpi_device_name(device), acpi_device_bid(device),
- (u32) ec->gpe_bit);
+ (u32) ec->common.gpe_bit);
if (!first_ec)
first_ec = device;
@@ -822,7 +1261,7 @@ acpi_ec_remove (
struct acpi_device *device,
int type)
{
- struct acpi_ec *ec;
+ union acpi_ec *ec = NULL;
ACPI_FUNCTION_TRACE("acpi_ec_remove");
@@ -844,7 +1283,7 @@ acpi_ec_io_ports (
struct acpi_resource *resource,
void *context)
{
- struct acpi_ec *ec = (struct acpi_ec *) context;
+ union acpi_ec *ec = (union acpi_ec *) context;
struct acpi_generic_address *addr;
if (resource->id != ACPI_RSTYPE_IO) {
@@ -856,10 +1295,10 @@ acpi_ec_io_ports (
* the second address region returned is the status/command
* port.
*/
- if (ec->data_addr.register_bit_width == 0) {
- addr = &ec->data_addr;
- } else if (ec->command_addr.register_bit_width == 0) {
- addr = &ec->command_addr;
+ if (ec->common.data_addr.register_bit_width == 0) {
+ addr = &ec->common.data_addr;
+ } else if (ec->common.command_addr.register_bit_width == 0) {
+ addr = &ec->common.command_addr;
} else {
return AE_CTRL_TERMINATE;
}
@@ -877,8 +1316,8 @@ static int
acpi_ec_start (
struct acpi_device *device)
{
- acpi_status status;
- struct acpi_ec *ec;
+ acpi_status status = AE_OK;
+ union acpi_ec *ec = NULL;
ACPI_FUNCTION_TRACE("acpi_ec_start");
@@ -893,35 +1332,36 @@ acpi_ec_start (
/*
* Get I/O port addresses. Convert to GAS format.
*/
- status = acpi_walk_resources(ec->handle, METHOD_NAME__CRS,
+ status = acpi_walk_resources(ec->common.handle, METHOD_NAME__CRS,
acpi_ec_io_ports, ec);
- if (ACPI_FAILURE(status) || ec->command_addr.register_bit_width == 0) {
+ if (ACPI_FAILURE(status) || ec->common.command_addr.register_bit_width == 0) {
ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Error getting I/O port addresses"));
return_VALUE(-ENODEV);
}
- ec->status_addr = ec->command_addr;
+ ec->common.status_addr = ec->common.command_addr;
ACPI_DEBUG_PRINT((ACPI_DB_INFO, "gpe=0x%02x, ports=0x%2x,0x%2x\n",
- (u32) ec->gpe_bit, (u32) ec->command_addr.address,
- (u32) ec->data_addr.address));
+ (u32) ec->common.gpe_bit, (u32) ec->common.command_addr.address,
+ (u32) ec->common.data_addr.address));
+
/*
* Install GPE handler
*/
- status = acpi_install_gpe_handler(NULL, ec->gpe_bit,
+ status = acpi_install_gpe_handler(NULL, ec->common.gpe_bit,
ACPI_GPE_EDGE_TRIGGERED, &acpi_ec_gpe_handler, ec);
if (ACPI_FAILURE(status)) {
return_VALUE(-ENODEV);
}
- acpi_set_gpe_type (NULL, ec->gpe_bit, ACPI_GPE_TYPE_RUNTIME);
- acpi_enable_gpe (NULL, ec->gpe_bit, ACPI_NOT_ISR);
+ acpi_set_gpe_type (NULL, ec->common.gpe_bit, ACPI_GPE_TYPE_RUNTIME);
+ acpi_enable_gpe (NULL, ec->common.gpe_bit, ACPI_NOT_ISR);
- status = acpi_install_address_space_handler (ec->handle,
+ status = acpi_install_address_space_handler (ec->common.handle,
ACPI_ADR_SPACE_EC, &acpi_ec_space_handler,
&acpi_ec_space_setup, ec);
if (ACPI_FAILURE(status)) {
- acpi_remove_gpe_handler(NULL, ec->gpe_bit, &acpi_ec_gpe_handler);
+ acpi_remove_gpe_handler(NULL, ec->common.gpe_bit, &acpi_ec_gpe_handler);
return_VALUE(-ENODEV);
}
@@ -934,8 +1374,8 @@ acpi_ec_stop (
struct acpi_device *device,
int type)
{
- acpi_status status;
- struct acpi_ec *ec;
+ acpi_status status = AE_OK;
+ union acpi_ec *ec = NULL;
ACPI_FUNCTION_TRACE("acpi_ec_stop");
@@ -944,12 +1384,12 @@ acpi_ec_stop (
ec = acpi_driver_data(device);
- status = acpi_remove_address_space_handler(ec->handle,
+ status = acpi_remove_address_space_handler(ec->common.handle,
ACPI_ADR_SPACE_EC, &acpi_ec_space_handler);
if (ACPI_FAILURE(status))
return_VALUE(-ENODEV);
- status = acpi_remove_gpe_handler(NULL, ec->gpe_bit, &acpi_ec_gpe_handler);
+ status = acpi_remove_gpe_handler(NULL, ec->common.gpe_bit, &acpi_ec_gpe_handler);
if (ACPI_FAILURE(status))
return_VALUE(-ENODEV);
@@ -963,26 +1403,76 @@ acpi_fake_ecdt_callback (
void *context,
void **retval)
{
+
+ if (acpi_ec_polling_mode)
+ return acpi_fake_ecdt_polling_callback(handle,
+ Level, context, retval);
+ else
+ return acpi_fake_ecdt_burst_callback(handle,
+ Level, context, retval);
+}
+
+static acpi_status __init
+acpi_fake_ecdt_polling_callback (
+ acpi_handle handle,
+ u32 Level,
+ void *context,
+ void **retval)
+{
+ acpi_status status;
+
+ status = acpi_walk_resources(handle, METHOD_NAME__CRS,
+ acpi_ec_io_ports, ec_ecdt);
+ if (ACPI_FAILURE(status))
+ return status;
+ ec_ecdt->common.status_addr = ec_ecdt->common.command_addr;
+
+ ec_ecdt->common.uid = -1;
+ acpi_evaluate_integer(handle, "_UID", NULL, &ec_ecdt->common.uid);
+
+ status = acpi_evaluate_integer(handle, "_GPE", NULL, &ec_ecdt->common.gpe_bit);
+ if (ACPI_FAILURE(status))
+ return status;
+ spin_lock_init(&ec_ecdt->polling.lock);
+ ec_ecdt->common.global_lock = TRUE;
+ ec_ecdt->common.handle = handle;
+
+ printk(KERN_INFO PREFIX "GPE=0x%02x, ports=0x%2x, 0x%2x\n",
+ (u32) ec_ecdt->common.gpe_bit, (u32) ec_ecdt->common.command_addr.address,
+ (u32) ec_ecdt->common.data_addr.address);
+
+ return AE_CTRL_TERMINATE;
+}
+
+static acpi_status __init
+acpi_fake_ecdt_burst_callback (
+ acpi_handle handle,
+ u32 Level,
+ void *context,
+ void **retval)
+{
acpi_status status;
+ init_MUTEX(&ec_ecdt->burst.sem);
+ init_waitqueue_head(&ec_ecdt->burst.wait);
status = acpi_walk_resources(handle, METHOD_NAME__CRS,
acpi_ec_io_ports, ec_ecdt);
if (ACPI_FAILURE(status))
return status;
- ec_ecdt->status_addr = ec_ecdt->command_addr;
+ ec_ecdt->common.status_addr = ec_ecdt->common.command_addr;
- ec_ecdt->uid = -1;
- acpi_evaluate_integer(handle, "_UID", NULL, &ec_ecdt->uid);
+ ec_ecdt->common.uid = -1;
+ acpi_evaluate_integer(handle, "_UID", NULL, &ec_ecdt->common.uid);
- status = acpi_evaluate_integer(handle, "_GPE", NULL, &ec_ecdt->gpe_bit);
+ status = acpi_evaluate_integer(handle, "_GPE", NULL, &ec_ecdt->common.gpe_bit);
if (ACPI_FAILURE(status))
return status;
- ec_ecdt->global_lock = TRUE;
- ec_ecdt->handle = handle;
+ ec_ecdt->common.global_lock = TRUE;
+ ec_ecdt->common.handle = handle;
printk(KERN_INFO PREFIX "GPE=0x%02x, ports=0x%2x, 0x%2x\n",
- (u32) ec_ecdt->gpe_bit, (u32) ec_ecdt->command_addr.address,
- (u32) ec_ecdt->data_addr.address);
+ (u32) ec_ecdt->common.gpe_bit, (u32) ec_ecdt->common.command_addr.address,
+ (u32) ec_ecdt->common.data_addr.address);
return AE_CTRL_TERMINATE;
}
@@ -1005,12 +1495,12 @@ acpi_ec_fake_ecdt(void)
printk(KERN_INFO PREFIX "Try to make an fake ECDT\n");
- ec_ecdt = kmalloc(sizeof(struct acpi_ec), GFP_KERNEL);
+ ec_ecdt = kmalloc(sizeof(union acpi_ec), GFP_KERNEL);
if (!ec_ecdt) {
ret = -ENOMEM;
goto error;
}
- memset(ec_ecdt, 0, sizeof(struct acpi_ec));
+ memset(ec_ecdt, 0, sizeof(union acpi_ec));
status = acpi_get_devices (ACPI_EC_HID,
acpi_fake_ecdt_callback,
@@ -1031,6 +1521,60 @@ error:
static int __init
acpi_ec_get_real_ecdt(void)
{
+ if (acpi_ec_polling_mode)
+ return acpi_ec_polling_get_real_ecdt();
+ else
+ return acpi_ec_burst_get_real_ecdt();
+}
+
+static int __init
+acpi_ec_polling_get_real_ecdt(void)
+{
+ acpi_status status;
+ struct acpi_table_ecdt *ecdt_ptr;
+
+ status = acpi_get_firmware_table("ECDT", 1, ACPI_LOGICAL_ADDRESSING,
+ (struct acpi_table_header **) &ecdt_ptr);
+ if (ACPI_FAILURE(status))
+ return -ENODEV;
+
+ printk(KERN_INFO PREFIX "Found ECDT\n");
+
+ /*
+ * Generate a temporary ec context to use until the namespace is scanned
+ */
+ ec_ecdt = kmalloc(sizeof(union acpi_ec), GFP_KERNEL);
+ if (!ec_ecdt)
+ return -ENOMEM;
+ memset(ec_ecdt, 0, sizeof(union acpi_ec));
+
+ ec_ecdt->common.command_addr = ecdt_ptr->ec_control;
+ ec_ecdt->common.status_addr = ecdt_ptr->ec_control;
+ ec_ecdt->common.data_addr = ecdt_ptr->ec_data;
+ ec_ecdt->common.gpe_bit = ecdt_ptr->gpe_bit;
+ spin_lock_init(&ec_ecdt->polling.lock);
+ /* use the GL just to be safe */
+ ec_ecdt->common.global_lock = TRUE;
+ ec_ecdt->common.uid = ecdt_ptr->uid;
+
+ status = acpi_get_handle(NULL, ecdt_ptr->ec_id, &ec_ecdt->common.handle);
+ if (ACPI_FAILURE(status)) {
+ goto error;
+ }
+
+ return 0;
+error:
+ printk(KERN_ERR PREFIX "Could not use ECDT\n");
+ kfree(ec_ecdt);
+ ec_ecdt = NULL;
+
+ return -ENODEV;
+}
+
+
+static int __init
+acpi_ec_burst_get_real_ecdt(void)
+{
acpi_status status;
struct acpi_table_ecdt *ecdt_ptr;
@@ -1044,22 +1588,22 @@ acpi_ec_get_real_ecdt(void)
/*
* Generate a temporary ec context to use until the namespace is scanned
*/
- ec_ecdt = kmalloc(sizeof(struct acpi_ec), GFP_KERNEL);
+ ec_ecdt = kmalloc(sizeof(union acpi_ec), GFP_KERNEL);
if (!ec_ecdt)
return -ENOMEM;
- memset(ec_ecdt, 0, sizeof(struct acpi_ec));
-
- init_MUTEX(&ec_ecdt->sem);
- init_waitqueue_head(&ec_ecdt->wait);
- ec_ecdt->command_addr = ecdt_ptr->ec_control;
- ec_ecdt->status_addr = ecdt_ptr->ec_control;
- ec_ecdt->data_addr = ecdt_ptr->ec_data;
- ec_ecdt->gpe_bit = ecdt_ptr->gpe_bit;
+ memset(ec_ecdt, 0, sizeof(union acpi_ec));
+
+ init_MUTEX(&ec_ecdt->burst.sem);
+ init_waitqueue_head(&ec_ecdt->burst.wait);
+ ec_ecdt->common.command_addr = ecdt_ptr->ec_control;
+ ec_ecdt->common.status_addr = ecdt_ptr->ec_control;
+ ec_ecdt->common.data_addr = ecdt_ptr->ec_data;
+ ec_ecdt->common.gpe_bit = ecdt_ptr->gpe_bit;
/* use the GL just to be safe */
- ec_ecdt->global_lock = TRUE;
- ec_ecdt->uid = ecdt_ptr->uid;
+ ec_ecdt->common.global_lock = TRUE;
+ ec_ecdt->common.uid = ecdt_ptr->uid;
- status = acpi_get_handle(NULL, ecdt_ptr->ec_id, &ec_ecdt->handle);
+ status = acpi_get_handle(NULL, ecdt_ptr->ec_id, &ec_ecdt->common.handle);
if (ACPI_FAILURE(status)) {
goto error;
}
@@ -1092,20 +1636,20 @@ acpi_ec_ecdt_probe (void)
/*
* Install GPE handler
*/
- status = acpi_install_gpe_handler(NULL, ec_ecdt->gpe_bit,
+ status = acpi_install_gpe_handler(NULL, ec_ecdt->common.gpe_bit,
ACPI_GPE_EDGE_TRIGGERED, &acpi_ec_gpe_handler,
ec_ecdt);
if (ACPI_FAILURE(status)) {
goto error;
}
- acpi_set_gpe_type (NULL, ec_ecdt->gpe_bit, ACPI_GPE_TYPE_RUNTIME);
- acpi_enable_gpe (NULL, ec_ecdt->gpe_bit, ACPI_NOT_ISR);
+ acpi_set_gpe_type (NULL, ec_ecdt->common.gpe_bit, ACPI_GPE_TYPE_RUNTIME);
+ acpi_enable_gpe (NULL, ec_ecdt->common.gpe_bit, ACPI_NOT_ISR);
status = acpi_install_address_space_handler (ACPI_ROOT_OBJECT,
ACPI_ADR_SPACE_EC, &acpi_ec_space_handler,
&acpi_ec_space_setup, ec_ecdt);
if (ACPI_FAILURE(status)) {
- acpi_remove_gpe_handler(NULL, ec_ecdt->gpe_bit,
+ acpi_remove_gpe_handler(NULL, ec_ecdt->common.gpe_bit,
&acpi_ec_gpe_handler);
goto error;
}
@@ -1123,7 +1667,7 @@ error:
static int __init acpi_ec_init (void)
{
- int result;
+ int result = 0;
ACPI_FUNCTION_TRACE("acpi_ec_init");
@@ -1167,3 +1711,10 @@ static int __init acpi_fake_ecdt_setup(char *str)
return 0;
}
__setup("acpi_fake_ecdt", acpi_fake_ecdt_setup);
+static int __init acpi_ec_set_polling_mode(char *str)
+{
+ acpi_ec_polling_mode = EC_POLLING;
+ acpi_ec_driver.ops.add = acpi_ec_polling_add;
+ return 0;
+}
+__setup("ec_polling", acpi_ec_set_polling_mode);
diff --git a/drivers/acpi/pci_irq.c b/drivers/acpi/pci_irq.c
index d1f42b972821..bb973d2109a1 100644
--- a/drivers/acpi/pci_irq.c
+++ b/drivers/acpi/pci_irq.c
@@ -269,7 +269,51 @@ acpi_pci_irq_del_prt (int segment, int bus)
/* --------------------------------------------------------------------------
PCI Interrupt Routing Support
-------------------------------------------------------------------------- */
+typedef int (*irq_lookup_func)(struct acpi_prt_entry *, int *, int *, char **);
+static int
+acpi_pci_allocate_irq(struct acpi_prt_entry *entry,
+ int *edge_level,
+ int *active_high_low,
+ char **link)
+{
+ int irq;
+
+ ACPI_FUNCTION_TRACE("acpi_pci_allocate_irq");
+
+ if (entry->link.handle) {
+ irq = acpi_pci_link_allocate_irq(entry->link.handle,
+ entry->link.index, edge_level, active_high_low, link);
+ if (irq < 0) {
+ ACPI_DEBUG_PRINT((ACPI_DB_WARN, "Invalid IRQ link routing entry\n"));
+ return_VALUE(-1);
+ }
+ } else {
+ irq = entry->link.index;
+ *edge_level = ACPI_LEVEL_SENSITIVE;
+ *active_high_low = ACPI_ACTIVE_LOW;
+ }
+
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Found IRQ %d\n", irq));
+ return_VALUE(irq);
+}
+
+static int
+acpi_pci_free_irq(struct acpi_prt_entry *entry,
+ int *edge_level,
+ int *active_high_low,
+ char **link)
+{
+ int irq;
+
+ ACPI_FUNCTION_TRACE("acpi_pci_free_irq");
+ if (entry->link.handle) {
+ irq = acpi_pci_link_free_irq(entry->link.handle);
+ } else {
+ irq = entry->link.index;
+ }
+ return_VALUE(irq);
+}
/*
* acpi_pci_irq_lookup
* success: return IRQ >= 0
@@ -282,12 +326,13 @@ acpi_pci_irq_lookup (
int pin,
int *edge_level,
int *active_high_low,
- char **link)
+ char **link,
+ irq_lookup_func func)
{
struct acpi_prt_entry *entry = NULL;
int segment = pci_domain_nr(bus);
int bus_nr = bus->number;
- int irq;
+ int ret;
ACPI_FUNCTION_TRACE("acpi_pci_irq_lookup");
@@ -301,22 +346,8 @@ acpi_pci_irq_lookup (
return_VALUE(-1);
}
- if (entry->link.handle) {
- irq = acpi_pci_link_get_irq(entry->link.handle,
- entry->link.index, edge_level, active_high_low, link);
- if (irq < 0) {
- ACPI_DEBUG_PRINT((ACPI_DB_WARN, "Invalid IRQ link routing entry\n"));
- return_VALUE(-1);
- }
- } else {
- irq = entry->link.index;
- *edge_level = ACPI_LEVEL_SENSITIVE;
- *active_high_low = ACPI_ACTIVE_LOW;
- }
-
- ACPI_DEBUG_PRINT((ACPI_DB_INFO, "Found IRQ %d\n", irq));
-
- return_VALUE(irq);
+ ret = func(entry, edge_level, active_high_low, link);
+ return_VALUE(ret);
}
/*
@@ -330,7 +361,8 @@ acpi_pci_irq_derive (
int pin,
int *edge_level,
int *active_high_low,
- char **link)
+ char **link,
+ irq_lookup_func func)
{
struct pci_dev *bridge = dev;
int irq = -1;
@@ -363,7 +395,7 @@ acpi_pci_irq_derive (
}
irq = acpi_pci_irq_lookup(bridge->bus, PCI_SLOT(bridge->devfn),
- pin, edge_level, active_high_low, link);
+ pin, edge_level, active_high_low, link, func);
}
if (irq < 0) {
@@ -415,7 +447,7 @@ acpi_pci_irq_enable (
* values override any BIOS-assigned IRQs set during boot.
*/
irq = acpi_pci_irq_lookup(dev->bus, PCI_SLOT(dev->devfn), pin,
- &edge_level, &active_high_low, &link);
+ &edge_level, &active_high_low, &link, acpi_pci_allocate_irq);
/*
* If no PRT entry was found, we'll try to derive an IRQ from the
@@ -423,7 +455,7 @@ acpi_pci_irq_enable (
*/
if (irq < 0)
irq = acpi_pci_irq_derive(dev, pin, &edge_level,
- &active_high_low, &link);
+ &active_high_low, &link, acpi_pci_allocate_irq);
/*
* No IRQ known to the ACPI subsystem - maybe the BIOS /
@@ -462,7 +494,9 @@ acpi_pci_irq_enable (
EXPORT_SYMBOL(acpi_pci_irq_enable);
-#ifdef CONFIG_ACPI_DEALLOCATE_IRQ
+/* FIXME: implement x86/x86_64 version */
+void __attribute__((weak)) acpi_unregister_gsi(u32 i) {}
+
void
acpi_pci_irq_disable (
struct pci_dev *dev)
@@ -489,14 +523,14 @@ acpi_pci_irq_disable (
* First we check the PCI IRQ routing table (PRT) for an IRQ.
*/
gsi = acpi_pci_irq_lookup(dev->bus, PCI_SLOT(dev->devfn), pin,
- &edge_level, &active_high_low, NULL);
+ &edge_level, &active_high_low, NULL, acpi_pci_free_irq);
/*
* If no PRT entry was found, we'll try to derive an IRQ from the
* device's parent bridge.
*/
if (gsi < 0)
gsi = acpi_pci_irq_derive(dev, pin,
- &edge_level, &active_high_low, NULL);
+ &edge_level, &active_high_low, NULL, acpi_pci_free_irq);
if (gsi < 0)
return_VOID;
@@ -512,4 +546,3 @@ acpi_pci_irq_disable (
return_VOID;
}
-#endif /* CONFIG_ACPI_DEALLOCATE_IRQ */
diff --git a/drivers/acpi/pci_link.c b/drivers/acpi/pci_link.c
index 6ad0e77df9b3..6a29610edc11 100644
--- a/drivers/acpi/pci_link.c
+++ b/drivers/acpi/pci_link.c
@@ -68,6 +68,10 @@ static struct acpi_driver acpi_pci_link_driver = {
},
};
+/*
+ * If a link is initialized, we never change its active and initialized
+ * later even the link is disable. Instead, we just repick the active irq
+ */
struct acpi_pci_link_irq {
u8 active; /* Current IRQ */
u8 edge_level; /* All IRQs */
@@ -76,8 +80,7 @@ struct acpi_pci_link_irq {
u8 possible_count;
u8 possible[ACPI_PCI_LINK_MAX_POSSIBLE];
u8 initialized:1;
- u8 suspend_resume:1;
- u8 reserved:6;
+ u8 reserved:7;
};
struct acpi_pci_link {
@@ -85,12 +88,14 @@ struct acpi_pci_link {
struct acpi_device *device;
acpi_handle handle;
struct acpi_pci_link_irq irq;
+ int refcnt;
};
static struct {
int count;
struct list_head entries;
} acpi_link;
+DECLARE_MUTEX(acpi_link_lock);
/* --------------------------------------------------------------------------
@@ -532,12 +537,12 @@ static int acpi_pci_link_allocate(
ACPI_FUNCTION_TRACE("acpi_pci_link_allocate");
- if (link->irq.suspend_resume) {
- acpi_pci_link_set(link, link->irq.active);
- link->irq.suspend_resume = 0;
- }
- if (link->irq.initialized)
+ if (link->irq.initialized) {
+ if (link->refcnt == 0)
+ /* This means the link is disabled but initialized */
+ acpi_pci_link_set(link, link->irq.active);
return_VALUE(0);
+ }
/*
* search for active IRQ in list of possible IRQs.
@@ -596,13 +601,13 @@ static int acpi_pci_link_allocate(
}
/*
- * acpi_pci_link_get_irq
+ * acpi_pci_link_allocate_irq
* success: return IRQ >= 0
* failure: return -1
*/
int
-acpi_pci_link_get_irq (
+acpi_pci_link_allocate_irq (
acpi_handle handle,
int index,
int *edge_level,
@@ -613,7 +618,7 @@ acpi_pci_link_get_irq (
struct acpi_device *device = NULL;
struct acpi_pci_link *link = NULL;
- ACPI_FUNCTION_TRACE("acpi_pci_link_get_irq");
+ ACPI_FUNCTION_TRACE("acpi_pci_link_allocate_irq");
result = acpi_bus_get_device(handle, &device);
if (result) {
@@ -633,21 +638,70 @@ acpi_pci_link_get_irq (
return_VALUE(-1);
}
- if (acpi_pci_link_allocate(link))
+ down(&acpi_link_lock);
+ if (acpi_pci_link_allocate(link)) {
+ up(&acpi_link_lock);
return_VALUE(-1);
+ }
if (!link->irq.active) {
+ up(&acpi_link_lock);
ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Link active IRQ is 0!\n"));
return_VALUE(-1);
}
+ link->refcnt ++;
+ up(&acpi_link_lock);
if (edge_level) *edge_level = link->irq.edge_level;
if (active_high_low) *active_high_low = link->irq.active_high_low;
if (name) *name = acpi_device_bid(link->device);
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO,
+ "Link %s is referenced\n", acpi_device_bid(link->device)));
return_VALUE(link->irq.active);
}
+/*
+ * We don't change link's irq information here. After it is reenabled, we
+ * continue use the info
+ */
+int
+acpi_pci_link_free_irq(acpi_handle handle)
+{
+ struct acpi_device *device = NULL;
+ struct acpi_pci_link *link = NULL;
+ acpi_status result;
+
+ ACPI_FUNCTION_TRACE("acpi_pci_link_free_irq");
+
+ result = acpi_bus_get_device(handle, &device);
+ if (result) {
+ ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Invalid link device\n"));
+ return_VALUE(-1);
+ }
+ link = (struct acpi_pci_link *) acpi_driver_data(device);
+ if (!link) {
+ ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Invalid link context\n"));
+ return_VALUE(-1);
+ }
+
+ down(&acpi_link_lock);
+ if (!link->irq.initialized) {
+ up(&acpi_link_lock);
+ ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Link isn't initialized\n"));
+ return_VALUE(-1);
+ }
+
+ link->refcnt --;
+ ACPI_DEBUG_PRINT((ACPI_DB_INFO,
+ "Link %s is dereferenced\n", acpi_device_bid(link->device)));
+
+ if (link->refcnt == 0) {
+ acpi_ut_evaluate_object(link->handle, "_DIS", 0, NULL);
+ }
+ up(&acpi_link_lock);
+ return_VALUE(link->irq.active);
+}
/* --------------------------------------------------------------------------
Driver Interface
-------------------------------------------------------------------------- */
@@ -677,6 +731,7 @@ acpi_pci_link_add (
strcpy(acpi_device_class(device), ACPI_PCI_LINK_CLASS);
acpi_driver_data(device) = link;
+ down(&acpi_link_lock);
result = acpi_pci_link_get_possible(link);
if (result)
goto end;
@@ -712,6 +767,7 @@ acpi_pci_link_add (
end:
/* disable all links -- to be activated on use */
acpi_ut_evaluate_object(link->handle, "_DIS", 0, NULL);
+ up(&acpi_link_lock);
if (result)
kfree(link);
@@ -726,19 +782,32 @@ irqrouter_suspend(
{
struct list_head *node = NULL;
struct acpi_pci_link *link = NULL;
+ int ret = 0;
ACPI_FUNCTION_TRACE("irqrouter_suspend");
list_for_each(node, &acpi_link.entries) {
link = list_entry(node, struct acpi_pci_link, node);
if (!link) {
- ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Invalid link context\n"));
+ ACPI_DEBUG_PRINT((ACPI_DB_ERROR,
+ "Invalid link context\n"));
continue;
}
- if (link->irq.active && link->irq.initialized)
- link->irq.suspend_resume = 1;
+ if (link->irq.initialized && link->refcnt != 0
+ /* We ignore legacy IDE device irq */
+ && link->irq.active != 14 && link->irq.active !=15) {
+ printk(KERN_WARNING PREFIX
+ "%d drivers with interrupt %d neglected to call"
+ " pci_disable_device at .suspend\n",
+ link->refcnt,
+ link->irq.active);
+ printk(KERN_WARNING PREFIX
+ "Fix the driver, or rmmod before suspend\n");
+ link->refcnt = 0;
+ ret = -EINVAL;
+ }
}
- return_VALUE(0);
+ return_VALUE(ret);
}
@@ -756,8 +825,9 @@ acpi_pci_link_remove (
link = (struct acpi_pci_link *) acpi_driver_data(device);
- /* TBD: Acquire/release lock */
+ down(&acpi_link_lock);
list_del(&link->node);
+ up(&acpi_link_lock);
kfree(link);
@@ -849,6 +919,7 @@ int __init acpi_irq_balance_set(char *str)
__setup("acpi_irq_balance", acpi_irq_balance_set);
+/* FIXME: we will remove this interface after all drivers call pci_disable_device */
static struct sysdev_class irqrouter_sysdev_class = {
set_kset_name("irqrouter"),
.suspend = irqrouter_suspend,
diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c
index 893b074e3d1a..af271d994f15 100644
--- a/drivers/acpi/processor_idle.c
+++ b/drivers/acpi/processor_idle.c
@@ -81,30 +81,33 @@ module_param(bm_history, uint, 0644);
*
* To skip this limit, boot/load with a large max_cstate limit.
*/
-static int no_c2c3(struct dmi_system_id *id)
+static int set_max_cstate(struct dmi_system_id *id)
{
if (max_cstate > ACPI_PROCESSOR_MAX_POWER)
return 0;
- printk(KERN_NOTICE PREFIX "%s detected - C2,C3 disabled."
+ printk(KERN_NOTICE PREFIX "%s detected - %s disabled."
" Override with \"processor.max_cstate=%d\"\n", id->ident,
+ ((int)id->driver_data == 1)? "C2,C3":"C3",
ACPI_PROCESSOR_MAX_POWER + 1);
- max_cstate = 1;
+ max_cstate = (int)id->driver_data;
return 0;
}
-
-
static struct dmi_system_id __initdata processor_power_dmi_table[] = {
- { no_c2c3, "IBM ThinkPad R40e", {
+ { set_max_cstate, "IBM ThinkPad R40e", {
DMI_MATCH(DMI_BIOS_VENDOR,"IBM"),
- DMI_MATCH(DMI_BIOS_VERSION,"1SET60WW") }},
- { no_c2c3, "Medion 41700", {
+ DMI_MATCH(DMI_BIOS_VERSION,"1SET60WW") }, (void*)1},
+ { set_max_cstate, "Medion 41700", {
+ DMI_MATCH(DMI_BIOS_VENDOR,"Phoenix Technologies LTD"),
+ DMI_MATCH(DMI_BIOS_VERSION,"R01-A1J") }, (void*)1},
+ { set_max_cstate, "Clevo 5600D", {
DMI_MATCH(DMI_BIOS_VENDOR,"Phoenix Technologies LTD"),
- DMI_MATCH(DMI_BIOS_VERSION,"R01-A1J") }},
+ DMI_MATCH(DMI_BIOS_VERSION,"SHE845M0.86C.0013.D.0302131307") },
+ (void*)2},
{},
};
@@ -549,7 +552,8 @@ static int acpi_processor_get_power_info_default_c1 (struct acpi_processor *pr)
ACPI_FUNCTION_TRACE("acpi_processor_get_power_info_default_c1");
for (i = 0; i < ACPI_PROCESSOR_MAX_POWER; i++)
- memset(pr->power.states, 0, sizeof(struct acpi_processor_cx));
+ memset(&(pr->power.states[i]), 0,
+ sizeof(struct acpi_processor_cx));
/* if info is obtained from pblk/fadt, type equals state */
pr->power.states[ACPI_STATE_C1].type = ACPI_STATE_C1;
@@ -580,7 +584,8 @@ static int acpi_processor_get_power_info_cst (struct acpi_processor *pr)
pr->power.count = 0;
for (i = 0; i < ACPI_PROCESSOR_MAX_POWER; i++)
- memset(pr->power.states, 0, sizeof(struct acpi_processor_cx));
+ memset(&(pr->power.states[i]), 0,
+ sizeof(struct acpi_processor_cx));
status = acpi_evaluate_object(pr->handle, "_CST", NULL, &buffer);
if (ACPI_FAILURE(status)) {
@@ -763,7 +768,6 @@ static void acpi_processor_power_verify_c3(
}
if (pr->flags.bm_check) {
- printk("Disabling BM access before entering C3\n");
/* bus mastering control is necessary */
if (!pr->flags.bm_control) {
ACPI_DEBUG_PRINT((ACPI_DB_INFO,
@@ -771,7 +775,6 @@ static void acpi_processor_power_verify_c3(
return_VOID;
}
} else {
- printk("Invalidating cache before entering C3\n");
/*
* WBINVD should be set in fadt, for C3 state to be
* supported on when bm_check is not required.
@@ -842,7 +845,7 @@ static int acpi_processor_get_power_info (
result = acpi_processor_get_power_info_cst(pr);
if ((result) || (acpi_processor_power_verify(pr) < 2)) {
result = acpi_processor_get_power_info_fadt(pr);
- if (result)
+ if ((result) || (acpi_processor_power_verify(pr) < 2))
result = acpi_processor_get_power_info_default_c1(pr);
}