linux_dsm_epyc7002/drivers/acpi/acpica/evxface.c

1080 lines
30 KiB
C
Raw Normal View History

// SPDX-License-Identifier: BSD-3-Clause OR GPL-2.0
/******************************************************************************
*
* Module Name: evxface - External interfaces for ACPI events
*
* Copyright (C) 2000 - 2018, Intel Corp.
*
*****************************************************************************/
#define EXPORT_ACPI_INTERFACES
#include <acpi/acpi.h>
#include "accommon.h"
#include "acnamesp.h"
#include "acevents.h"
#include "acinterp.h"
#define _COMPONENT ACPI_EVENTS
ACPI_MODULE_NAME("evxface")
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
#if (!ACPI_REDUCED_HARDWARE)
/* Local prototypes */
static acpi_status
acpi_ev_install_gpe_handler(acpi_handle gpe_device,
u32 gpe_number,
u32 type,
u8 is_raw_handler,
acpi_gpe_handler address, void *context);
#endif
/*******************************************************************************
*
* FUNCTION: acpi_install_notify_handler
*
* PARAMETERS: device - The device for which notifies will be handled
* handler_type - The type of handler:
* ACPI_SYSTEM_NOTIFY: System Handler (00-7F)
* ACPI_DEVICE_NOTIFY: Device Handler (80-FF)
* ACPI_ALL_NOTIFY: Both System and Device
* handler - Address of the handler
* context - Value passed to the handler on each GPE
*
* RETURN: Status
*
* DESCRIPTION: Install a handler for notifications on an ACPI Device,
* thermal_zone, or Processor object.
*
* NOTES: The Root namespace object may have only one handler for each
* type of notify (System/Device). Device/Thermal/Processor objects
* may have one device notify handler, and multiple system notify
* handlers.
*
******************************************************************************/
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
acpi_status
acpi_install_notify_handler(acpi_handle device,
u32 handler_type,
acpi_notify_handler handler, void *context)
{
struct acpi_namespace_node *node =
ACPI_CAST_PTR(struct acpi_namespace_node, device);
union acpi_operand_object *obj_desc;
union acpi_operand_object *handler_obj;
acpi_status status;
u32 i;
ACPI: ACPICA 20060421 Removed a device initialization optimization introduced in 20051216 where the _STA method was not run unless an _INI was also present for the same device. This optimization could cause problems because it could allow _INI methods to be run within a not-present device subtree (If a not-present device had no _INI, _STA would not be run, the not-present status would not be discovered, and the children of the device would be incorrectly traversed.) Implemented a new _STA optimization where namespace subtrees that do not contain _INI are identified and ignored during device initialization. Selectively running _STA can significantly improve boot time on large machines (with assistance from Len Brown.) Implemented support for the device initialization case where the returned _STA flags indicate a device not-present but functioning. In this case, _INI is not run, but the device children are examined for presence, as per the ACPI specification. Implemented an additional change to the IndexField support in order to conform to MS behavior. The value written to the Index Register is not simply a byte offset, it is a byte offset in units of the access width of the parent Index Field. (Fiodor Suietov) Defined and deployed a new OSL interface, acpi_os_validate_address(). This interface is called during the creation of all AML operation regions, and allows the host OS to exert control over what addresses it will allow the AML code to access. Operation Regions whose addresses are disallowed will cause a runtime exception when they are actually accessed (will not affect or abort table loading.) Defined and deployed a new OSL interface, acpi_os_validate_interface(). This interface allows the host OS to match the various "optional" interface/behavior strings for the _OSI predefined control method as appropriate (with assistance from Bjorn Helgaas.) Restructured and corrected various problems in the exception handling code paths within DsCallControlMethod and DsTerminateControlMethod in dsmethod (with assistance from Takayoshi Kochi.) Modified the Linux source converter to ignore quoted string literals while converting identifiers from mixed to lower case. This will correct problems with the disassembler and other areas where such strings must not be modified. The ACPI_FUNCTION_* macros no longer require quotes around the function name. This allows the Linux source converter to convert the names, now that the converter ignores quoted strings. Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-04-22 04:15:00 +07:00
ACPI_FUNCTION_TRACE(acpi_install_notify_handler);
/* Parameter validation */
if ((!device) || (!handler) || (!handler_type) ||
(handler_type > ACPI_MAX_NOTIFY_HANDLER_TYPE)) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_NAMESPACE);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/*
* Root Object:
* Registering a notify handler on the root object indicates that the
* caller wishes to receive notifications for all objects. Note that
* only one global handler can be registered per notify type.
* Ensure that a handler is not already installed.
*/
if (device == ACPI_ROOT_OBJECT) {
for (i = 0; i < ACPI_NUM_NOTIFY_TYPES; i++) {
if (handler_type & (i + 1)) {
if (acpi_gbl_global_notify[i].handler) {
status = AE_ALREADY_EXISTS;
goto unlock_and_exit;
}
acpi_gbl_global_notify[i].handler = handler;
acpi_gbl_global_notify[i].context = context;
}
}
goto unlock_and_exit; /* Global notify handler installed, all done */
}
/*
* All Other Objects:
* Caller will only receive notifications specific to the target
* object. Note that only certain object types are allowed to
* receive notifications.
*/
/* Are Notifies allowed on this object? */
if (!acpi_ev_is_notify_object(node)) {
status = AE_TYPE;
goto unlock_and_exit;
}
/* Check for an existing internal object, might not exist */
obj_desc = acpi_ns_get_attached_object(node);
if (!obj_desc) {
/* Create a new object */
obj_desc = acpi_ut_create_internal_object(node->type);
if (!obj_desc) {
status = AE_NO_MEMORY;
goto unlock_and_exit;
}
/* Attach new object to the Node, remove local reference */
status = acpi_ns_attach_object(device, obj_desc, node->type);
acpi_ut_remove_reference(obj_desc);
if (ACPI_FAILURE(status)) {
goto unlock_and_exit;
}
}
/* Ensure that the handler is not already installed in the lists */
for (i = 0; i < ACPI_NUM_NOTIFY_TYPES; i++) {
if (handler_type & (i + 1)) {
handler_obj = obj_desc->common_notify.notify_list[i];
while (handler_obj) {
if (handler_obj->notify.handler == handler) {
status = AE_ALREADY_EXISTS;
goto unlock_and_exit;
}
handler_obj = handler_obj->notify.next[i];
}
}
}
/* Create and populate a new notify handler object */
handler_obj = acpi_ut_create_internal_object(ACPI_TYPE_LOCAL_NOTIFY);
if (!handler_obj) {
status = AE_NO_MEMORY;
goto unlock_and_exit;
}
handler_obj->notify.node = node;
handler_obj->notify.handler_type = handler_type;
handler_obj->notify.handler = handler;
handler_obj->notify.context = context;
/* Install the handler at the list head(s) */
for (i = 0; i < ACPI_NUM_NOTIFY_TYPES; i++) {
if (handler_type & (i + 1)) {
handler_obj->notify.next[i] =
obj_desc->common_notify.notify_list[i];
obj_desc->common_notify.notify_list[i] = handler_obj;
}
}
/* Add an extra reference if handler was installed in both lists */
if (handler_type == ACPI_ALL_NOTIFY) {
acpi_ut_add_reference(handler_obj);
}
unlock_and_exit:
(void)acpi_ut_release_mutex(ACPI_MTX_NAMESPACE);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_notify_handler)
/*******************************************************************************
*
* FUNCTION: acpi_remove_notify_handler
*
* PARAMETERS: device - The device for which the handler is installed
* handler_type - The type of handler:
* ACPI_SYSTEM_NOTIFY: System Handler (00-7F)
* ACPI_DEVICE_NOTIFY: Device Handler (80-FF)
* ACPI_ALL_NOTIFY: Both System and Device
* handler - Address of the handler
*
* RETURN: Status
*
* DESCRIPTION: Remove a handler for notifies on an ACPI device
*
******************************************************************************/
acpi_status
acpi_remove_notify_handler(acpi_handle device,
u32 handler_type, acpi_notify_handler handler)
{
struct acpi_namespace_node *node =
ACPI_CAST_PTR(struct acpi_namespace_node, device);
union acpi_operand_object *obj_desc;
union acpi_operand_object *handler_obj;
union acpi_operand_object *previous_handler_obj;
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
acpi_status status = AE_OK;
u32 i;
ACPI: ACPICA 20060421 Removed a device initialization optimization introduced in 20051216 where the _STA method was not run unless an _INI was also present for the same device. This optimization could cause problems because it could allow _INI methods to be run within a not-present device subtree (If a not-present device had no _INI, _STA would not be run, the not-present status would not be discovered, and the children of the device would be incorrectly traversed.) Implemented a new _STA optimization where namespace subtrees that do not contain _INI are identified and ignored during device initialization. Selectively running _STA can significantly improve boot time on large machines (with assistance from Len Brown.) Implemented support for the device initialization case where the returned _STA flags indicate a device not-present but functioning. In this case, _INI is not run, but the device children are examined for presence, as per the ACPI specification. Implemented an additional change to the IndexField support in order to conform to MS behavior. The value written to the Index Register is not simply a byte offset, it is a byte offset in units of the access width of the parent Index Field. (Fiodor Suietov) Defined and deployed a new OSL interface, acpi_os_validate_address(). This interface is called during the creation of all AML operation regions, and allows the host OS to exert control over what addresses it will allow the AML code to access. Operation Regions whose addresses are disallowed will cause a runtime exception when they are actually accessed (will not affect or abort table loading.) Defined and deployed a new OSL interface, acpi_os_validate_interface(). This interface allows the host OS to match the various "optional" interface/behavior strings for the _OSI predefined control method as appropriate (with assistance from Bjorn Helgaas.) Restructured and corrected various problems in the exception handling code paths within DsCallControlMethod and DsTerminateControlMethod in dsmethod (with assistance from Takayoshi Kochi.) Modified the Linux source converter to ignore quoted string literals while converting identifiers from mixed to lower case. This will correct problems with the disassembler and other areas where such strings must not be modified. The ACPI_FUNCTION_* macros no longer require quotes around the function name. This allows the Linux source converter to convert the names, now that the converter ignores quoted strings. Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-04-22 04:15:00 +07:00
ACPI_FUNCTION_TRACE(acpi_remove_notify_handler);
/* Parameter validation */
if ((!device) || (!handler) || (!handler_type) ||
(handler_type > ACPI_MAX_NOTIFY_HANDLER_TYPE)) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
ACPICA: Cleanup indentation to reduce differences between Linux and ACPICA. This is a cosmetic patch only. Comparison of the resulting binary showed only line number differences. This patch does not affect the generation of the Linux binary. This patch decreases 210 lines of 20121018 divergence.diff. The ACPICA source codes uses a totally different indentation style from the Linux to be compatible with other users (operating systems or BIOS). Indentation differences are critical to the release automation. There are two causes related to the "indentation" that are affecting the release automation: 1. The ACPICA -> Linux release process is: ACPICA source -- acpisrc - hierarchy - indent -> linuxized ACPICA source -- diff -> linuxized ACPICA patch (x) -- human intervention -> linuxized ACPICA patch (o) Where 'x' means "cannot be directly applied to the Linux" 'o' means "can be directly applied to the Linux" Different "indent" version or "indent" options used in the "indent" step will lead to different divergences. The version of "indent" used for the current release process is: GNU indent 2.2.11 The options of "indent" used for the current release process is: -npro -kr -i8 -ts8 -sob -l80 -ss -ncs 2. Manual indentation prettifying work in the Linux side will also harm the automatically generated linuxized ACPICA patches, making them impossible to apply directly. This patch fixes source code differences caused by the two causes so that the "human intervention" can be reduced in the future. Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2012-12-19 12:37:15 +07:00
/* Root Object. Global handlers are removed here */
if (device == ACPI_ROOT_OBJECT) {
for (i = 0; i < ACPI_NUM_NOTIFY_TYPES; i++) {
if (handler_type & (i + 1)) {
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
status =
acpi_ut_acquire_mutex(ACPI_MTX_NAMESPACE);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
if (!acpi_gbl_global_notify[i].handler ||
(acpi_gbl_global_notify[i].handler !=
handler)) {
status = AE_NOT_EXIST;
goto unlock_and_exit;
}
ACPI_DEBUG_PRINT((ACPI_DB_INFO,
"Removing global notify handler\n"));
acpi_gbl_global_notify[i].handler = NULL;
acpi_gbl_global_notify[i].context = NULL;
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
(void)acpi_ut_release_mutex(ACPI_MTX_NAMESPACE);
/* Make sure all deferred notify tasks are completed */
acpi_os_wait_events_complete();
}
}
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
return_ACPI_STATUS(AE_OK);
}
/* All other objects: Are Notifies allowed on this object? */
if (!acpi_ev_is_notify_object(node)) {
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
return_ACPI_STATUS(AE_TYPE);
}
/* Must have an existing internal object */
obj_desc = acpi_ns_get_attached_object(node);
if (!obj_desc) {
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
return_ACPI_STATUS(AE_NOT_EXIST);
}
/* Internal object exists. Find the handler and remove it */
for (i = 0; i < ACPI_NUM_NOTIFY_TYPES; i++) {
if (handler_type & (i + 1)) {
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
status = acpi_ut_acquire_mutex(ACPI_MTX_NAMESPACE);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
handler_obj = obj_desc->common_notify.notify_list[i];
previous_handler_obj = NULL;
/* Attempt to find the handler in the handler list */
ACPI: ACPICA 20060707 Added the ACPI_PACKED_POINTERS_NOT_SUPPORTED macro to support C compilers that do not allow the initialization of address pointers within packed structures - even though the hardware itself may support misaligned transfers. Some of the debug data structures are packed by default to minimize size. Added an error message for the case where acpi_os_get_thread_id() returns zero. A non-zero value is required by the core ACPICA code to ensure the proper operation of AML mutexes and recursive control methods. The DSDT is now the only ACPI table that determines whether the AML interpreter is in 32-bit or 64-bit mode. Not really a functional change, but the hooks for per-table 32/64 switching have been removed from the code. A clarification to the ACPI specification is forthcoming in ACPI 3.0B. Fixed a possible leak of an Owner ID in the error path of tbinstal.c acpi_tb_init_table_descriptor() and migrated all table OwnerID deletion to a single place in acpi_tb_uninstall_table() to correct possible leaks when using the acpi_tb_delete_tables_by_type() interface (with assistance from Lance Ortiz.) Fixed a problem with Serialized control methods where the semaphore associated with the method could be over-signaled after multiple method invocations. Fixed two issues with the locking of the internal namespace data structure. Both the Unload() operator and acpi_unload_table() interface now lock the namespace during the namespace deletion associated with the table unload (with assistance from Linn Crosetto.) Fixed problem reports (Valery Podrezov) integrated: - Eliminate unnecessary memory allocation for CreateXxxxField http://bugzilla.kernel.org/show_bug.cgi?id=5426 Fixed problem reports (Fiodor Suietov) integrated: - Incomplete cleanup branches in AcpiTbGetTableRsdt (BZ 369) - On Address Space handler deletion, needless deactivation call (BZ 374) - AcpiRemoveAddressSpaceHandler: validate Device handle parameter (BZ 375) - Possible memory leak, Notify sub-objects of Processor, Power, ThermalZone (BZ 376) - AcpiRemoveAddressSpaceHandler: validate Handler parameter (BZ 378) - Minimum Length of RSDT should be validated (BZ 379) - AcpiRemoveNotifyHandler: return AE_NOT_EXIST if Processor Obj has no Handler (BZ (380) - AcpiUnloadTable: return AE_NOT_EXIST if no table of specified type loaded (BZ 381) Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-07-08 07:44:38 +07:00
while (handler_obj &&
(handler_obj->notify.handler != handler)) {
previous_handler_obj = handler_obj;
handler_obj = handler_obj->notify.next[i];
}
if (!handler_obj) {
status = AE_NOT_EXIST;
ACPI: ACPICA 20060707 Added the ACPI_PACKED_POINTERS_NOT_SUPPORTED macro to support C compilers that do not allow the initialization of address pointers within packed structures - even though the hardware itself may support misaligned transfers. Some of the debug data structures are packed by default to minimize size. Added an error message for the case where acpi_os_get_thread_id() returns zero. A non-zero value is required by the core ACPICA code to ensure the proper operation of AML mutexes and recursive control methods. The DSDT is now the only ACPI table that determines whether the AML interpreter is in 32-bit or 64-bit mode. Not really a functional change, but the hooks for per-table 32/64 switching have been removed from the code. A clarification to the ACPI specification is forthcoming in ACPI 3.0B. Fixed a possible leak of an Owner ID in the error path of tbinstal.c acpi_tb_init_table_descriptor() and migrated all table OwnerID deletion to a single place in acpi_tb_uninstall_table() to correct possible leaks when using the acpi_tb_delete_tables_by_type() interface (with assistance from Lance Ortiz.) Fixed a problem with Serialized control methods where the semaphore associated with the method could be over-signaled after multiple method invocations. Fixed two issues with the locking of the internal namespace data structure. Both the Unload() operator and acpi_unload_table() interface now lock the namespace during the namespace deletion associated with the table unload (with assistance from Linn Crosetto.) Fixed problem reports (Valery Podrezov) integrated: - Eliminate unnecessary memory allocation for CreateXxxxField http://bugzilla.kernel.org/show_bug.cgi?id=5426 Fixed problem reports (Fiodor Suietov) integrated: - Incomplete cleanup branches in AcpiTbGetTableRsdt (BZ 369) - On Address Space handler deletion, needless deactivation call (BZ 374) - AcpiRemoveAddressSpaceHandler: validate Device handle parameter (BZ 375) - Possible memory leak, Notify sub-objects of Processor, Power, ThermalZone (BZ 376) - AcpiRemoveAddressSpaceHandler: validate Handler parameter (BZ 378) - Minimum Length of RSDT should be validated (BZ 379) - AcpiRemoveNotifyHandler: return AE_NOT_EXIST if Processor Obj has no Handler (BZ (380) - AcpiUnloadTable: return AE_NOT_EXIST if no table of specified type loaded (BZ 381) Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-07-08 07:44:38 +07:00
goto unlock_and_exit;
}
/* Remove the handler object from the list */
if (previous_handler_obj) { /* Handler is not at the list head */
previous_handler_obj->notify.next[i] =
handler_obj->notify.next[i];
} else { /* Handler is at the list head */
ACPI: ACPICA 20060707 Added the ACPI_PACKED_POINTERS_NOT_SUPPORTED macro to support C compilers that do not allow the initialization of address pointers within packed structures - even though the hardware itself may support misaligned transfers. Some of the debug data structures are packed by default to minimize size. Added an error message for the case where acpi_os_get_thread_id() returns zero. A non-zero value is required by the core ACPICA code to ensure the proper operation of AML mutexes and recursive control methods. The DSDT is now the only ACPI table that determines whether the AML interpreter is in 32-bit or 64-bit mode. Not really a functional change, but the hooks for per-table 32/64 switching have been removed from the code. A clarification to the ACPI specification is forthcoming in ACPI 3.0B. Fixed a possible leak of an Owner ID in the error path of tbinstal.c acpi_tb_init_table_descriptor() and migrated all table OwnerID deletion to a single place in acpi_tb_uninstall_table() to correct possible leaks when using the acpi_tb_delete_tables_by_type() interface (with assistance from Lance Ortiz.) Fixed a problem with Serialized control methods where the semaphore associated with the method could be over-signaled after multiple method invocations. Fixed two issues with the locking of the internal namespace data structure. Both the Unload() operator and acpi_unload_table() interface now lock the namespace during the namespace deletion associated with the table unload (with assistance from Linn Crosetto.) Fixed problem reports (Valery Podrezov) integrated: - Eliminate unnecessary memory allocation for CreateXxxxField http://bugzilla.kernel.org/show_bug.cgi?id=5426 Fixed problem reports (Fiodor Suietov) integrated: - Incomplete cleanup branches in AcpiTbGetTableRsdt (BZ 369) - On Address Space handler deletion, needless deactivation call (BZ 374) - AcpiRemoveAddressSpaceHandler: validate Device handle parameter (BZ 375) - Possible memory leak, Notify sub-objects of Processor, Power, ThermalZone (BZ 376) - AcpiRemoveAddressSpaceHandler: validate Handler parameter (BZ 378) - Minimum Length of RSDT should be validated (BZ 379) - AcpiRemoveNotifyHandler: return AE_NOT_EXIST if Processor Obj has no Handler (BZ (380) - AcpiUnloadTable: return AE_NOT_EXIST if no table of specified type loaded (BZ 381) Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-07-08 07:44:38 +07:00
obj_desc->common_notify.notify_list[i] =
handler_obj->notify.next[i];
}
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
(void)acpi_ut_release_mutex(ACPI_MTX_NAMESPACE);
/* Make sure all deferred notify tasks are completed */
acpi_os_wait_events_complete();
acpi_ut_remove_reference(handler_obj);
}
}
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
return_ACPI_STATUS(status);
unlock_and_exit:
(void)acpi_ut_release_mutex(ACPI_MTX_NAMESPACE);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_remove_notify_handler)
/*******************************************************************************
*
* FUNCTION: acpi_install_exception_handler
*
* PARAMETERS: handler - Pointer to the handler function for the
* event
*
* RETURN: Status
*
* DESCRIPTION: Saves the pointer to the handler function
*
******************************************************************************/
#ifdef ACPI_FUTURE_USAGE
acpi_status acpi_install_exception_handler(acpi_exception_handler handler)
{
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_install_exception_handler);
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/* Don't allow two handlers. */
if (acpi_gbl_exception_handler) {
status = AE_ALREADY_EXISTS;
goto cleanup;
}
/* Install the handler */
acpi_gbl_exception_handler = handler;
cleanup:
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_exception_handler)
ACPI: Enable build of AML interpreter debugger This patch enables ACPICA debugger files using a configurable CONFIG_ACPI_DEBUGGER configuration item. Those debugger related code that was originally masked as ACPI_FUTURE_USAGE now gets unmasked. Necessary OSL stubs are also added in this patch: 1. acpi_os_readable(): This should be arch specific in Linux, while this patch doesn't introduce real implementation and a complex mechanism to allow architecture specific acpi_os_readable() to be implemented to validate the address. It may be done by future commits. 2. acpi_os_get_line(): This is used to obtain debugger command input. This patch only introduces a simple KDB concept example in it and the example should be co-working with the code implemented in acpi_os_printf(). Since this KDB example won't be compiled unless ENABLE_DEBUGGER is defined and it seems Linux has already stopped to use ENABLE_DEBUGGER, thus do not expect it can work properly. This patch also cleans up all other ACPI_FUTURE_USAGE surroundings accordingly. 1. Since linkage error can be automatically detected, declaration in the headers needn't be surrounded by ACPI_FUTURE_USAGE. So only the following separate exported fuction bodies are masked by this macro (other exported fucntions may have already been masked at entire module level via drivers/acpi/acpica/Makefile): acpi_install_exception_handler() acpi_subsystem_status() acpi_get_system_info() acpi_get_statistics() acpi_install_initialization_handler() 2. Since strip can automatically zap the no-user functions, functions that are not marked with ACPI_EXPORT_SYMBOL() needn't get surrounded by ACPI_FUTURE_USAGE. So the following function which is not used by Linux kernel now won't get surrounded by this macro: acpi_ps_get_name() Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-10-19 09:25:56 +07:00
#endif
#if (!ACPI_REDUCED_HARDWARE)
/*******************************************************************************
*
* FUNCTION: acpi_install_sci_handler
*
* PARAMETERS: address - Address of the handler
* context - Value passed to the handler on each SCI
*
* RETURN: Status
*
* DESCRIPTION: Install a handler for a System Control Interrupt.
*
******************************************************************************/
acpi_status acpi_install_sci_handler(acpi_sci_handler address, void *context)
{
struct acpi_sci_handler_info *new_sci_handler;
struct acpi_sci_handler_info *sci_handler;
acpi_cpu_flags flags;
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_install_sci_handler);
if (!address) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
/* Allocate and init a handler object */
new_sci_handler = ACPI_ALLOCATE(sizeof(struct acpi_sci_handler_info));
if (!new_sci_handler) {
return_ACPI_STATUS(AE_NO_MEMORY);
}
new_sci_handler->address = address;
new_sci_handler->context = context;
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
goto exit;
}
/* Lock list during installation */
flags = acpi_os_acquire_lock(acpi_gbl_gpe_lock);
sci_handler = acpi_gbl_sci_handler_list;
/* Ensure handler does not already exist */
while (sci_handler) {
if (address == sci_handler->address) {
status = AE_ALREADY_EXISTS;
goto unlock_and_exit;
}
sci_handler = sci_handler->next;
}
/* Install the new handler into the global list (at head) */
new_sci_handler->next = acpi_gbl_sci_handler_list;
acpi_gbl_sci_handler_list = new_sci_handler;
unlock_and_exit:
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
exit:
if (ACPI_FAILURE(status)) {
ACPI_FREE(new_sci_handler);
}
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_sci_handler)
/*******************************************************************************
*
* FUNCTION: acpi_remove_sci_handler
*
* PARAMETERS: address - Address of the handler
*
* RETURN: Status
*
* DESCRIPTION: Remove a handler for a System Control Interrupt.
*
******************************************************************************/
acpi_status acpi_remove_sci_handler(acpi_sci_handler address)
{
struct acpi_sci_handler_info *prev_sci_handler;
struct acpi_sci_handler_info *next_sci_handler;
acpi_cpu_flags flags;
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_remove_sci_handler);
if (!address) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/* Remove the SCI handler with lock */
flags = acpi_os_acquire_lock(acpi_gbl_gpe_lock);
prev_sci_handler = NULL;
next_sci_handler = acpi_gbl_sci_handler_list;
while (next_sci_handler) {
if (next_sci_handler->address == address) {
/* Unlink and free the SCI handler info block */
if (prev_sci_handler) {
prev_sci_handler->next = next_sci_handler->next;
} else {
acpi_gbl_sci_handler_list =
next_sci_handler->next;
}
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
ACPI_FREE(next_sci_handler);
goto unlock_and_exit;
}
prev_sci_handler = next_sci_handler;
next_sci_handler = next_sci_handler->next;
}
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
status = AE_NOT_EXIST;
unlock_and_exit:
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_remove_sci_handler)
/*******************************************************************************
*
* FUNCTION: acpi_install_global_event_handler
*
* PARAMETERS: handler - Pointer to the global event handler function
* context - Value passed to the handler on each event
*
* RETURN: Status
*
* DESCRIPTION: Saves the pointer to the handler function. The global handler
* is invoked upon each incoming GPE and Fixed Event. It is
* invoked at interrupt level at the time of the event dispatch.
* Can be used to update event counters, etc.
*
******************************************************************************/
acpi_status
acpi_install_global_event_handler(acpi_gbl_event_handler handler, void *context)
{
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_install_global_event_handler);
/* Parameter validation */
if (!handler) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/* Don't allow two handlers. */
if (acpi_gbl_global_event_handler) {
status = AE_ALREADY_EXISTS;
goto cleanup;
}
acpi_gbl_global_event_handler = handler;
acpi_gbl_global_event_handler_context = context;
cleanup:
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_global_event_handler)
/*******************************************************************************
*
* FUNCTION: acpi_install_fixed_event_handler
*
* PARAMETERS: event - Event type to enable.
* handler - Pointer to the handler function for the
* event
* context - Value passed to the handler on each GPE
*
* RETURN: Status
*
* DESCRIPTION: Saves the pointer to the handler function and then enables the
* event.
*
******************************************************************************/
acpi_status
acpi_install_fixed_event_handler(u32 event,
acpi_event_handler handler, void *context)
{
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_install_fixed_event_handler);
/* Parameter validation */
if (event > ACPI_EVENT_MAX) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/* Do not allow multiple handlers */
if (acpi_gbl_fixed_event_handlers[event].handler) {
status = AE_ALREADY_EXISTS;
goto cleanup;
}
/* Install the handler before enabling the event */
acpi_gbl_fixed_event_handlers[event].handler = handler;
acpi_gbl_fixed_event_handlers[event].context = context;
status = acpi_clear_event(event);
if (ACPI_SUCCESS(status))
status = acpi_enable_event(event, 0);
if (ACPI_FAILURE(status)) {
ACPI_WARNING((AE_INFO,
"Could not enable fixed event - %s (%u)",
acpi_ut_get_event_name(event), event));
/* Remove the handler */
acpi_gbl_fixed_event_handlers[event].handler = NULL;
acpi_gbl_fixed_event_handlers[event].context = NULL;
} else {
ACPI_DEBUG_PRINT((ACPI_DB_INFO,
"Enabled fixed event %s (%X), Handler=%p\n",
acpi_ut_get_event_name(event), event,
handler));
}
cleanup:
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_fixed_event_handler)
/*******************************************************************************
*
* FUNCTION: acpi_remove_fixed_event_handler
*
* PARAMETERS: event - Event type to disable.
* handler - Address of the handler
*
* RETURN: Status
*
* DESCRIPTION: Disables the event and unregisters the event handler.
*
******************************************************************************/
acpi_status
acpi_remove_fixed_event_handler(u32 event, acpi_event_handler handler)
{
acpi_status status = AE_OK;
ACPI_FUNCTION_TRACE(acpi_remove_fixed_event_handler);
/* Parameter validation */
if (event > ACPI_EVENT_MAX) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/* Disable the event before removing the handler */
status = acpi_disable_event(event, 0);
/* Always Remove the handler */
acpi_gbl_fixed_event_handlers[event].handler = NULL;
acpi_gbl_fixed_event_handlers[event].context = NULL;
if (ACPI_FAILURE(status)) {
ACPI_WARNING((AE_INFO,
"Could not disable fixed event - %s (%u)",
acpi_ut_get_event_name(event), event));
} else {
ACPI_DEBUG_PRINT((ACPI_DB_INFO,
"Disabled fixed event - %s (%X)\n",
acpi_ut_get_event_name(event), event));
}
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_remove_fixed_event_handler)
/*******************************************************************************
*
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
* FUNCTION: acpi_ev_install_gpe_handler
*
ACPICA 20050408 from Bob Moore Fixed three cases in the interpreter where an "index" argument to an ASL function was still (internally) 32 bits instead of the required 64 bits. This was the Index argument to the Index, Mid, and Match operators. The "strupr" function is now permanently local (acpi_ut_strupr), since this is not a POSIX-defined function and not present in most kernel-level C libraries. References to the C library strupr function have been removed from the headers. Completed the deployment of static functions/prototypes. All prototypes with the static attribute have been moved from the headers to the owning C file. ACPICA 20050329 from Bob Moore An error is now generated if an attempt is made to create a Buffer Field of length zero (A CreateField with a length operand of zero.) The interpreter now issues a warning whenever executable code at the module level is detected during ACPI table load. This will give some idea of the prevalence of this type of code. Implemented support for references to named objects (other than control methods) within package objects. Enhanced package object output for the debug object. Package objects are now completely dumped, showing all elements. Enhanced miscellaneous object output for the debug object. Any object can now be written to the debug object (for example, a device object can be written, and the type of the object will be displayed.) The "static" qualifier has been added to all local functions across the core subsystem. The number of "long" lines (> 80 chars) within the source has been significantly reduced, by about 1/3. Cleaned up all header files to ensure that all CA/iASL functions are prototyped (even static functions) and the formatting is consistent. Two new header files have been added, acopcode.h and acnames.h. Removed several obsolete functions that were no longer used. Signed-off-by: Len Brown <len.brown@intel.com>
2005-04-19 09:49:35 +07:00
* PARAMETERS: gpe_device - Namespace node for the GPE (NULL for FADT
* defined GPEs)
* gpe_number - The GPE number within the GPE block
* type - Whether this GPE should be treated as an
* edge- or level-triggered interrupt.
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
* is_raw_handler - Whether this GPE should be handled using
* the special GPE handler mode.
* address - Address of the handler
* context - Value passed to the handler on each GPE
*
* RETURN: Status
*
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
* DESCRIPTION: Internal function to install a handler for a General Purpose
* Event.
*
******************************************************************************/
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
static acpi_status
acpi_ev_install_gpe_handler(acpi_handle gpe_device,
u32 gpe_number,
u32 type,
u8 is_raw_handler,
acpi_gpe_handler address, void *context)
{
struct acpi_gpe_event_info *gpe_event_info;
struct acpi_gpe_handler_info *handler;
acpi_status status;
[ACPI] ACPICA 20060127 Implemented support in the Resource Manager to allow unresolved namestring references within resource package objects for the _PRT method. This support is in addition to the previously implemented unresolved reference support within the AML parser. If the interpreter slack mode is enabled (true on Linux unless acpi=strict), these unresolved references will be passed through to the caller as a NULL package entry. http://bugzilla.kernel.org/show_bug.cgi?id=5741 Implemented and deployed new macros and functions for error and warning messages across the subsystem. These macros are simpler and generate less code than their predecessors. The new macros ACPI_ERROR, ACPI_EXCEPTION, ACPI_WARNING, and ACPI_INFO replace the ACPI_REPORT_* macros. Implemented the acpi_cpu_flags type to simplify host OS integration of the Acquire/Release Lock OSL interfaces. Suggested by Steven Rostedt and Andrew Morton. Fixed a problem where Alias ASL operators are sometimes not correctly resolved. causing AE_AML_INTERNAL http://bugzilla.kernel.org/show_bug.cgi?id=5189 http://bugzilla.kernel.org/show_bug.cgi?id=5674 Fixed several problems with the implementation of the ConcatenateResTemplate ASL operator. As per the ACPI specification, zero length buffers are now treated as a single EndTag. One-length buffers always cause a fatal exception. Non-zero length buffers that do not end with a full 2-byte EndTag cause a fatal exception. Fixed a possible structure overwrite in the AcpiGetObjectInfo external interface. (With assistance from Thomas Renninger) Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-01-28 04:43:00 +07:00
acpi_cpu_flags flags;
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
ACPI_FUNCTION_TRACE(ev_install_gpe_handler);
/* Parameter validation */
if ((!address) || (type & ~ACPI_GPE_XRUPT_TYPE_MASK)) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
/* Allocate and init handler object (before lock) */
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
handler = ACPI_ALLOCATE_ZEROED(sizeof(struct acpi_gpe_handler_info));
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
if (!handler) {
status = AE_NO_MEMORY;
goto unlock_and_exit;
}
flags = acpi_os_acquire_lock(acpi_gbl_gpe_lock);
/* Ensure that we have a valid GPE number */
gpe_event_info = acpi_ev_get_gpe_event_info(gpe_device, gpe_number);
if (!gpe_event_info) {
status = AE_BAD_PARAMETER;
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
goto free_and_exit;
}
/* Make sure that there isn't a handler there already */
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
if ((ACPI_GPE_DISPATCH_TYPE(gpe_event_info->flags) ==
ACPI_GPE_DISPATCH_HANDLER) ||
(ACPI_GPE_DISPATCH_TYPE(gpe_event_info->flags) ==
ACPI_GPE_DISPATCH_RAW_HANDLER)) {
status = AE_ALREADY_EXISTS;
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
goto free_and_exit;
}
handler->address = address;
handler->context = context;
handler->method_node = gpe_event_info->dispatch.method_node;
ACPICA: Cleanup indentation to reduce differences between Linux and ACPICA. This is a cosmetic patch only. Comparison of the resulting binary showed only line number differences. This patch does not affect the generation of the Linux binary. This patch decreases 210 lines of 20121018 divergence.diff. The ACPICA source codes uses a totally different indentation style from the Linux to be compatible with other users (operating systems or BIOS). Indentation differences are critical to the release automation. There are two causes related to the "indentation" that are affecting the release automation: 1. The ACPICA -> Linux release process is: ACPICA source -- acpisrc - hierarchy - indent -> linuxized ACPICA source -- diff -> linuxized ACPICA patch (x) -- human intervention -> linuxized ACPICA patch (o) Where 'x' means "cannot be directly applied to the Linux" 'o' means "can be directly applied to the Linux" Different "indent" version or "indent" options used in the "indent" step will lead to different divergences. The version of "indent" used for the current release process is: GNU indent 2.2.11 The options of "indent" used for the current release process is: -npro -kr -i8 -ts8 -sob -l80 -ss -ncs 2. Manual indentation prettifying work in the Linux side will also harm the automatically generated linuxized ACPICA patches, making them impossible to apply directly. This patch fixes source code differences caused by the two causes so that the "human intervention" can be reduced in the future. Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2012-12-19 12:37:15 +07:00
handler->original_flags = (u8)(gpe_event_info->flags &
(ACPI_GPE_XRUPT_TYPE_MASK |
ACPI_GPE_DISPATCH_MASK));
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
/*
* If the GPE is associated with a method, it may have been enabled
* automatically during initialization, in which case it has to be
* disabled now to avoid spurious execution of the handler.
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
*/
ACPICA: Events: Cleanup GPE dispatcher type obtaining code ACPICA commit 7926d5ca9452c87f866938dcea8f12e1efb58f89 There is an issue in acpi_install_gpe_handler() and acpi_remove_gpe_handler(). The code to obtain the GPE dispatcher type from the Handler->original_flags is wrong: if (((Handler->original_flags & ACPI_GPE_DISPATCH_METHOD) || (Handler->original_flags & ACPI_GPE_DISPATCH_NOTIFY)) && ACPI_GPE_DISPATCH_NOTIFY is 0x03 and ACPI_GPE_DISPATCH_METHOD is 0x02, thus this statement is TRUE for the following dispatcher types: 0x01 (ACPI_GPE_DISPATCH_HANDLER): not expected 0x02 (ACPI_GPE_DISPATCH_METHOD): expected 0x03 (ACPI_GPE_DISPATCH_NOTIFY): expected There is no functional issue due to this because Handler->original_flags is only set in acpi_install_gpe_handler(), and an earlier checker has excluded the ACPI_GPE_DISPATCH_HANDLER: if ((gpe_event_info->Flags & ACPI_GPE_DISPATCH_MASK) == ACPI_GPE_DISPATCH_HANDLER) { Status = AE_ALREADY_EXISTS; goto free_and_exit; } ... Handler->original_flags = (u8) (gpe_event_info->Flags & (ACPI_GPE_XRUPT_TYPE_MASK | ACPI_GPE_DISPATCH_MASK)); We need to clean this up before modifying the GPE dispatcher type values. In order to prevent such issue from happening in the future, this patch introduces ACPI_GPE_DISPATCH_TYPE() macro to be used to obtain the GPE dispatcher types. Lv Zheng. Link: https://github.com/acpica/acpica/commit/7926d5ca Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: David E. Box <david.e.box@linux.intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 14:20:29 +07:00
if (((ACPI_GPE_DISPATCH_TYPE(handler->original_flags) ==
ACPI_GPE_DISPATCH_METHOD) ||
(ACPI_GPE_DISPATCH_TYPE(handler->original_flags) ==
ACPI_GPE_DISPATCH_NOTIFY)) && gpe_event_info->runtime_count) {
handler->originally_enabled = TRUE;
(void)acpi_ev_remove_gpe_reference(gpe_event_info);
/* Sanity check of original type against new type */
if (type !=
(u32)(gpe_event_info->flags & ACPI_GPE_XRUPT_TYPE_MASK)) {
ACPI_WARNING((AE_INFO,
"GPE type mismatch (level/edge)"));
}
}
/* Install the handler */
gpe_event_info->dispatch.handler = handler;
/* Setup up dispatch flags to indicate handler (vs. method/notify) */
gpe_event_info->flags &=
~(ACPI_GPE_XRUPT_TYPE_MASK | ACPI_GPE_DISPATCH_MASK);
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
gpe_event_info->flags |=
(u8)(type |
(is_raw_handler ? ACPI_GPE_DISPATCH_RAW_HANDLER :
ACPI_GPE_DISPATCH_HANDLER));
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
unlock_and_exit:
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
free_and_exit:
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
ACPI_FREE(handler);
goto unlock_and_exit;
}
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
/*******************************************************************************
*
* FUNCTION: acpi_install_gpe_handler
*
* PARAMETERS: gpe_device - Namespace node for the GPE (NULL for FADT
* defined GPEs)
* gpe_number - The GPE number within the GPE block
* type - Whether this GPE should be treated as an
* edge- or level-triggered interrupt.
* address - Address of the handler
* context - Value passed to the handler on each GPE
*
* RETURN: Status
*
* DESCRIPTION: Install a handler for a General Purpose Event.
*
******************************************************************************/
acpi_status
acpi_install_gpe_handler(acpi_handle gpe_device,
u32 gpe_number,
u32 type, acpi_gpe_handler address, void *context)
{
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_install_gpe_handler);
status = acpi_ev_install_gpe_handler(gpe_device, gpe_number, type,
FALSE, address, context);
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_gpe_handler)
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
/*******************************************************************************
*
* FUNCTION: acpi_install_gpe_raw_handler
*
* PARAMETERS: gpe_device - Namespace node for the GPE (NULL for FADT
* defined GPEs)
* gpe_number - The GPE number within the GPE block
* type - Whether this GPE should be treated as an
* edge- or level-triggered interrupt.
* address - Address of the handler
* context - Value passed to the handler on each GPE
*
* RETURN: Status
*
* DESCRIPTION: Install a handler for a General Purpose Event.
*
******************************************************************************/
acpi_status
acpi_install_gpe_raw_handler(acpi_handle gpe_device,
u32 gpe_number,
u32 type, acpi_gpe_handler address, void *context)
{
acpi_status status;
ACPI_FUNCTION_TRACE(acpi_install_gpe_raw_handler);
status = acpi_ev_install_gpe_handler(gpe_device, gpe_number, type,
TRUE, address, context);
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_install_gpe_raw_handler)
/*******************************************************************************
*
* FUNCTION: acpi_remove_gpe_handler
*
ACPICA 20050408 from Bob Moore Fixed three cases in the interpreter where an "index" argument to an ASL function was still (internally) 32 bits instead of the required 64 bits. This was the Index argument to the Index, Mid, and Match operators. The "strupr" function is now permanently local (acpi_ut_strupr), since this is not a POSIX-defined function and not present in most kernel-level C libraries. References to the C library strupr function have been removed from the headers. Completed the deployment of static functions/prototypes. All prototypes with the static attribute have been moved from the headers to the owning C file. ACPICA 20050329 from Bob Moore An error is now generated if an attempt is made to create a Buffer Field of length zero (A CreateField with a length operand of zero.) The interpreter now issues a warning whenever executable code at the module level is detected during ACPI table load. This will give some idea of the prevalence of this type of code. Implemented support for references to named objects (other than control methods) within package objects. Enhanced package object output for the debug object. Package objects are now completely dumped, showing all elements. Enhanced miscellaneous object output for the debug object. Any object can now be written to the debug object (for example, a device object can be written, and the type of the object will be displayed.) The "static" qualifier has been added to all local functions across the core subsystem. The number of "long" lines (> 80 chars) within the source has been significantly reduced, by about 1/3. Cleaned up all header files to ensure that all CA/iASL functions are prototyped (even static functions) and the formatting is consistent. Two new header files have been added, acopcode.h and acnames.h. Removed several obsolete functions that were no longer used. Signed-off-by: Len Brown <len.brown@intel.com>
2005-04-19 09:49:35 +07:00
* PARAMETERS: gpe_device - Namespace node for the GPE (NULL for FADT
* defined GPEs)
* gpe_number - The event to remove a handler
* address - Address of the handler
*
* RETURN: Status
*
* DESCRIPTION: Remove a handler for a General Purpose acpi_event.
*
******************************************************************************/
acpi_status
acpi_remove_gpe_handler(acpi_handle gpe_device,
u32 gpe_number, acpi_gpe_handler address)
{
struct acpi_gpe_event_info *gpe_event_info;
struct acpi_gpe_handler_info *handler;
acpi_status status;
[ACPI] ACPICA 20060127 Implemented support in the Resource Manager to allow unresolved namestring references within resource package objects for the _PRT method. This support is in addition to the previously implemented unresolved reference support within the AML parser. If the interpreter slack mode is enabled (true on Linux unless acpi=strict), these unresolved references will be passed through to the caller as a NULL package entry. http://bugzilla.kernel.org/show_bug.cgi?id=5741 Implemented and deployed new macros and functions for error and warning messages across the subsystem. These macros are simpler and generate less code than their predecessors. The new macros ACPI_ERROR, ACPI_EXCEPTION, ACPI_WARNING, and ACPI_INFO replace the ACPI_REPORT_* macros. Implemented the acpi_cpu_flags type to simplify host OS integration of the Acquire/Release Lock OSL interfaces. Suggested by Steven Rostedt and Andrew Morton. Fixed a problem where Alias ASL operators are sometimes not correctly resolved. causing AE_AML_INTERNAL http://bugzilla.kernel.org/show_bug.cgi?id=5189 http://bugzilla.kernel.org/show_bug.cgi?id=5674 Fixed several problems with the implementation of the ConcatenateResTemplate ASL operator. As per the ACPI specification, zero length buffers are now treated as a single EndTag. One-length buffers always cause a fatal exception. Non-zero length buffers that do not end with a full 2-byte EndTag cause a fatal exception. Fixed a possible structure overwrite in the AcpiGetObjectInfo external interface. (With assistance from Thomas Renninger) Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-01-28 04:43:00 +07:00
acpi_cpu_flags flags;
ACPI: ACPICA 20060421 Removed a device initialization optimization introduced in 20051216 where the _STA method was not run unless an _INI was also present for the same device. This optimization could cause problems because it could allow _INI methods to be run within a not-present device subtree (If a not-present device had no _INI, _STA would not be run, the not-present status would not be discovered, and the children of the device would be incorrectly traversed.) Implemented a new _STA optimization where namespace subtrees that do not contain _INI are identified and ignored during device initialization. Selectively running _STA can significantly improve boot time on large machines (with assistance from Len Brown.) Implemented support for the device initialization case where the returned _STA flags indicate a device not-present but functioning. In this case, _INI is not run, but the device children are examined for presence, as per the ACPI specification. Implemented an additional change to the IndexField support in order to conform to MS behavior. The value written to the Index Register is not simply a byte offset, it is a byte offset in units of the access width of the parent Index Field. (Fiodor Suietov) Defined and deployed a new OSL interface, acpi_os_validate_address(). This interface is called during the creation of all AML operation regions, and allows the host OS to exert control over what addresses it will allow the AML code to access. Operation Regions whose addresses are disallowed will cause a runtime exception when they are actually accessed (will not affect or abort table loading.) Defined and deployed a new OSL interface, acpi_os_validate_interface(). This interface allows the host OS to match the various "optional" interface/behavior strings for the _OSI predefined control method as appropriate (with assistance from Bjorn Helgaas.) Restructured and corrected various problems in the exception handling code paths within DsCallControlMethod and DsTerminateControlMethod in dsmethod (with assistance from Takayoshi Kochi.) Modified the Linux source converter to ignore quoted string literals while converting identifiers from mixed to lower case. This will correct problems with the disassembler and other areas where such strings must not be modified. The ACPI_FUNCTION_* macros no longer require quotes around the function name. This allows the Linux source converter to convert the names, now that the converter ignores quoted strings. Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2006-04-22 04:15:00 +07:00
ACPI_FUNCTION_TRACE(acpi_remove_gpe_handler);
/* Parameter validation */
if (!address) {
return_ACPI_STATUS(AE_BAD_PARAMETER);
}
status = acpi_ut_acquire_mutex(ACPI_MTX_EVENTS);
if (ACPI_FAILURE(status)) {
return_ACPI_STATUS(status);
}
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
flags = acpi_os_acquire_lock(acpi_gbl_gpe_lock);
/* Ensure that we have a valid GPE number */
gpe_event_info = acpi_ev_get_gpe_event_info(gpe_device, gpe_number);
if (!gpe_event_info) {
status = AE_BAD_PARAMETER;
goto unlock_and_exit;
}
/* Make sure that a handler is indeed installed */
ACPICA: Events: Introduce ACPI_GPE_DISPATCH_RAW_HANDLER to fix 2 issues for the current GPE APIs ACPICA commit 199cad16530a45aea2bec98e528866e20c5927e1 Since whether the GPE should be disabled/enabled/cleared should only be determined by the GPE driver's state machine: 1. GPE should be disabled if the driver wants to switch to the GPE polling mode when a GPE storm condition is indicated and should be enabled if the driver wants to switch back to the GPE interrupt mode when all of the storm conditions are cleared. The conditions should be protected by the driver's specific lock. 2. GPE should be enabled if the driver has accepted more than one request and should be disabled if the driver has completed all of the requests. The request count should be protected by the driver's specific lock. 3. GPE should be cleared either when the driver is about to handle an edge triggered GPE or when the driver has completed to handle a level triggered GPE. The handling code should be protected by the driver's specific lock. Thus the GPE enabling/disabling/clearing operations are likely to be performed with the driver's specific lock held while we currently cannot do this. This is because: 1. We have the acpi_gbl_gpe_lock held before invoking the GPE driver's handler. Driver's specific lock is likely to be held inside of the handler, thus we can see some dead lock issues due to the reversed locking order or recursive locking. In order to solve such dead lock issues, we need to unlock the acpi_gbl_gpe_lock before invoking the handler. BZ 1100. 2. Since GPE disabling/enabling/clearing should be determined by the GPE driver's state machine, we shouldn't perform such operations inside of ACPICA for a GPE handler to mess up the driver's state machine. BZ 1101. Originally this patch includes a logic to flush GPE handlers, it is dropped due to the following reasons: 1. This is a different issue; 2. Linux OSL has fixed this by flushing SCI in acpi_os_wait_events_complete(). We will pick up this topic when the Linux OSL fix turns out to be not sufficient. Note that currently the internal operations and the acpi_gbl_gpe_lock are also used by ACPI_GPE_DISPATCH_METHOD and ACPI_GPE_DISPATCH_NOTIFY. In order not to introduce regressions, we add one ACPI_GPE_DISPATCH_RAW_HANDLER type to be distiguished from ACPI_GPE_DISPATCH_HANDLER. For which the acpi_gbl_gpe_lock is unlocked before invoking the GPE handler and the internal enabling/disabling operations are bypassed to allow drivers to perform them at a proper position using the GPE APIs and ACPI_GPE_DISPATCH_RAW_HANDLER users should invoke acpi_set_gpe() instead of acpi_enable_gpe()/acpi_disable_gpe() to bypass the internal GPE clearing code in acpi_enable_gpe(). Lv Zheng. Known issues: 1. Edge-triggered GPE lost for frequent enablings On some buggy silicon platforms, GPE enable line may not be directly wired to the GPE trigger line. In that case, when GPE enabling is frequently performed for edge-triggered GPEs, GPE status may stay set without being triggered. This patch may maginify this problem as it allows GPE enabling to be parallel performed during the process the GPEs are handled. This is an existing issue, because: 1. For task context: Current ACPI_GPE_DISPATCH_METHOD practices have proven that this isn't a real issue - we can re-enable edge-triggered GPE in a work queue where the GPE status bit might already be set. 2. For IRQ context: This can even happen when the GPE enabling occurs before returning from the GPE handler and after unlocking the GPE lock. Thus currently no code is included to protect this. Link: https://github.com/acpica/acpica/commit/199cad16 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 15:27:03 +07:00
if ((ACPI_GPE_DISPATCH_TYPE(gpe_event_info->flags) !=
ACPI_GPE_DISPATCH_HANDLER) &&
(ACPI_GPE_DISPATCH_TYPE(gpe_event_info->flags) !=
ACPI_GPE_DISPATCH_RAW_HANDLER)) {
status = AE_NOT_EXIST;
goto unlock_and_exit;
}
/* Make sure that the installed handler is the same */
if (gpe_event_info->dispatch.handler->address != address) {
status = AE_BAD_PARAMETER;
goto unlock_and_exit;
}
/* Remove the handler */
handler = gpe_event_info->dispatch.handler;
gpe_event_info->dispatch.handler = NULL;
/* Restore Method node (if any), set dispatch flags */
gpe_event_info->dispatch.method_node = handler->method_node;
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
gpe_event_info->flags &=
~(ACPI_GPE_XRUPT_TYPE_MASK | ACPI_GPE_DISPATCH_MASK);
gpe_event_info->flags |= handler->original_flags;
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
/*
* If the GPE was previously associated with a method and it was
* enabled, it should be enabled at this point to restore the
* post-initialization configuration.
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
*/
ACPICA: Events: Cleanup GPE dispatcher type obtaining code ACPICA commit 7926d5ca9452c87f866938dcea8f12e1efb58f89 There is an issue in acpi_install_gpe_handler() and acpi_remove_gpe_handler(). The code to obtain the GPE dispatcher type from the Handler->original_flags is wrong: if (((Handler->original_flags & ACPI_GPE_DISPATCH_METHOD) || (Handler->original_flags & ACPI_GPE_DISPATCH_NOTIFY)) && ACPI_GPE_DISPATCH_NOTIFY is 0x03 and ACPI_GPE_DISPATCH_METHOD is 0x02, thus this statement is TRUE for the following dispatcher types: 0x01 (ACPI_GPE_DISPATCH_HANDLER): not expected 0x02 (ACPI_GPE_DISPATCH_METHOD): expected 0x03 (ACPI_GPE_DISPATCH_NOTIFY): expected There is no functional issue due to this because Handler->original_flags is only set in acpi_install_gpe_handler(), and an earlier checker has excluded the ACPI_GPE_DISPATCH_HANDLER: if ((gpe_event_info->Flags & ACPI_GPE_DISPATCH_MASK) == ACPI_GPE_DISPATCH_HANDLER) { Status = AE_ALREADY_EXISTS; goto free_and_exit; } ... Handler->original_flags = (u8) (gpe_event_info->Flags & (ACPI_GPE_XRUPT_TYPE_MASK | ACPI_GPE_DISPATCH_MASK)); We need to clean this up before modifying the GPE dispatcher type values. In order to prevent such issue from happening in the future, this patch introduces ACPI_GPE_DISPATCH_TYPE() macro to be used to obtain the GPE dispatcher types. Lv Zheng. Link: https://github.com/acpica/acpica/commit/7926d5ca Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: David E. Box <david.e.box@linux.intel.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2015-02-05 14:20:29 +07:00
if (((ACPI_GPE_DISPATCH_TYPE(handler->original_flags) ==
ACPI_GPE_DISPATCH_METHOD) ||
(ACPI_GPE_DISPATCH_TYPE(handler->original_flags) ==
ACPI_GPE_DISPATCH_NOTIFY)) && handler->originally_enabled) {
(void)acpi_ev_add_gpe_reference(gpe_event_info);
if (ACPI_GPE_IS_POLLING_NEEDED(gpe_event_info)) {
/* Poll edge triggered GPEs to handle existing events */
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
(void)acpi_ev_detect_gpe(gpe_device, gpe_event_info,
gpe_number);
flags = acpi_os_acquire_lock(acpi_gbl_gpe_lock);
}
ACPICA: Cleanup indentation to reduce differences between Linux and ACPICA. This is a cosmetic patch only. Comparison of the resulting binary showed only line number differences. This patch does not affect the generation of the Linux binary. This patch decreases 210 lines of 20121018 divergence.diff. The ACPICA source codes uses a totally different indentation style from the Linux to be compatible with other users (operating systems or BIOS). Indentation differences are critical to the release automation. There are two causes related to the "indentation" that are affecting the release automation: 1. The ACPICA -> Linux release process is: ACPICA source -- acpisrc - hierarchy - indent -> linuxized ACPICA source -- diff -> linuxized ACPICA patch (x) -- human intervention -> linuxized ACPICA patch (o) Where 'x' means "cannot be directly applied to the Linux" 'o' means "can be directly applied to the Linux" Different "indent" version or "indent" options used in the "indent" step will lead to different divergences. The version of "indent" used for the current release process is: GNU indent 2.2.11 The options of "indent" used for the current release process is: -npro -kr -i8 -ts8 -sob -l80 -ss -ncs 2. Manual indentation prettifying work in the Linux side will also harm the automatically generated linuxized ACPICA patches, making them impossible to apply directly. This patch fixes source code differences caused by the two causes so that the "human intervention" can be reduced in the future. Signed-off-by: Lv Zheng <lv.zheng@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2012-12-19 12:37:15 +07:00
}
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
/* Make sure all deferred GPE tasks are completed */
acpi_os_wait_events_complete();
/* Now we can free the handler object */
ACPI_FREE(handler);
ACPICA: Update use of acpi_os_wait_events_complete interface. This patch cleans up all of the acpi_os_wait_events_complete() invocations to make it to be invoked inside of ACPICA in the way to accommodate Linux's work queue implementation. According to the report, current Linux kernel code is facing a boot time race issue in the acpi_remove_notify_handler(). This is because: Linux is using work queues to implement a deferred handler call environment while ACPICA expects OSPM to implement acpi_os_wait_events_complete() using wait queues. The position to invoke a "waiter" is not suitable for a "flusher" as new invocations can be scheduled after this position and before the deletion of the handler from its management container. Since the following commit has deleted acpi_os_wait_events_complete() parameters, it thus might not be possible for OSPM to achieve a safe removal using wait queues. This requires ACPICA to be changed accordingly to "flush" handlers rather than "wait" them to be drain up: Commit: 5ff986a2a9db11858247b71fe242fe17617229aa Date: Wed, 16 May 2012 13:36:07 -0700 Subject: Introduce acpi_os_wait_events_complete interface. This interface will block until asynchronous events like notifies and GPEs are complete. Within ACPICA, it is called before a notify or GPE handler is removed. ACPICA BZ 868. This patch fixes this issue by invoking acpi_os_wait_events_complete() in the way to "flush" things - it thus should be put to the position after handler is removed from its management container but before it is destructed. The technical concerns are: 1. MTX_NAMESPACE is used to protect things that acpi_os_wait_events_complete() might be waiting for, thus MTX_NAMESPACE must be unlocked before invoking acpi_os_wait_events_complete(). 2. MTX_NAMESPACE is also used to implement the serialization of acpi_install_notify_handler() and acpi_remove_notify_handler(). This patch changes this logic, thus if there are many acpi_install/remove_notify_handler() invoked in parallel, the acpi_os_wait_events_complete() might face the races which could cause it never running to an end. Normally this will require additional code to implement a separate locking facility which is not implemented due to 3. 3. Given ACPICA users will always invoke acpi_install_notify_handler() once during Linux module/device initialization and invoke acpi_remove_notify_handler() once during module/device finalization, problem stated in 2 will not happen in Linux environment due to the mutual exclusive module/device existence, this fix thus is sufficient. Same concerns can apply to acpi_install/remove_gpe_handler(). Reported and tested: Ronald Vink. Fixed: Lv Zheng. References: https://bugzilla.kernel.org/show_bug.cgi?id=60583 Signed-off-by: Lv Zheng <lv.zheng@intel.com> Reported-and-Tested-by: Ronald Vink <ronald.vink@boskalis.com> Signed-off-by: Bob Moore <robert.moore@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2014-04-04 11:37:51 +07:00
return_ACPI_STATUS(status);
ACPI / ACPICA: Fix reference counting problems with GPE handlers If a handler is installed for a GPE associated with an AML method and such that it cannot wake up the system from sleep states, the GPE remains enabled after the handler has been installed, although it should be disabled in that case to avoid spurious execution of the handler. Fix this issue by making acpi_install_gpe_handler() disable GPEs that were previously associated with AML methods and cannot wake up the system from sleep states. Analogously, make acpi_remove_gpe_handler() enable the GPEs that are associated with AML methods after their handlers have been removed and cannot wake up the system from sleep states. In addition to that, fix a code ordering issue in acpi_remove_gpe_handler() that renders the locking ineffective (ACPI_MTX_EVENTS is released temporarily in the middle of the routine to wait for the completion of events already in progress). For this purpose introduce acpi_raw_disable_gpe() and acpi_raw_enable_gpe() to be called with acpi_gbl_gpe_lock held and rework acpi_disable_gpe() and acpi_enable_gpe(), respectively, to use them. Also rework acpi_gpe_can_wake() to use acpi_raw_disable_gpe() instead of calling acpi_disable_gpe() after releasing the lock to avoid the possible theoretical race with acpi_install_gpe_handler(). Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: "Moore, Robert" <robert.moore@intel.com> Cc: Lin Ming <ming.m.lin@intel.com> Signed-off-by: Len Brown <len.brown@intel.com>
2010-08-04 04:55:14 +07:00
unlock_and_exit:
acpi_os_release_lock(acpi_gbl_gpe_lock, flags);
(void)acpi_ut_release_mutex(ACPI_MTX_EVENTS);
return_ACPI_STATUS(status);
}
ACPI_EXPORT_SYMBOL(acpi_remove_gpe_handler)
/*******************************************************************************
*
* FUNCTION: acpi_acquire_global_lock
*
* PARAMETERS: timeout - How long the caller is willing to wait
* handle - Where the handle to the lock is returned
ACPICA 20050408 from Bob Moore Fixed three cases in the interpreter where an "index" argument to an ASL function was still (internally) 32 bits instead of the required 64 bits. This was the Index argument to the Index, Mid, and Match operators. The "strupr" function is now permanently local (acpi_ut_strupr), since this is not a POSIX-defined function and not present in most kernel-level C libraries. References to the C library strupr function have been removed from the headers. Completed the deployment of static functions/prototypes. All prototypes with the static attribute have been moved from the headers to the owning C file. ACPICA 20050329 from Bob Moore An error is now generated if an attempt is made to create a Buffer Field of length zero (A CreateField with a length operand of zero.) The interpreter now issues a warning whenever executable code at the module level is detected during ACPI table load. This will give some idea of the prevalence of this type of code. Implemented support for references to named objects (other than control methods) within package objects. Enhanced package object output for the debug object. Package objects are now completely dumped, showing all elements. Enhanced miscellaneous object output for the debug object. Any object can now be written to the debug object (for example, a device object can be written, and the type of the object will be displayed.) The "static" qualifier has been added to all local functions across the core subsystem. The number of "long" lines (> 80 chars) within the source has been significantly reduced, by about 1/3. Cleaned up all header files to ensure that all CA/iASL functions are prototyped (even static functions) and the formatting is consistent. Two new header files have been added, acopcode.h and acnames.h. Removed several obsolete functions that were no longer used. Signed-off-by: Len Brown <len.brown@intel.com>
2005-04-19 09:49:35 +07:00
* (if acquired)
*
* RETURN: Status
*
* DESCRIPTION: Acquire the ACPI Global Lock
*
* Note: Allows callers with the same thread ID to acquire the global lock
* multiple times. In other words, externally, the behavior of the global lock
* is identical to an AML mutex. On the first acquire, a new handle is
* returned. On any subsequent calls to acquire by the same thread, the same
* handle is returned.
*
******************************************************************************/
acpi_status acpi_acquire_global_lock(u16 timeout, u32 *handle)
{
acpi_status status;
if (!handle) {
return (AE_BAD_PARAMETER);
}
/* Must lock interpreter to prevent race conditions */
acpi_ex_enter_interpreter();
status = acpi_ex_acquire_mutex_object(timeout,
acpi_gbl_global_lock_mutex,
acpi_os_get_thread_id());
if (ACPI_SUCCESS(status)) {
/* Return the global lock handle (updated in acpi_ev_acquire_global_lock) */
*handle = acpi_gbl_global_lock_handle;
}
acpi_ex_exit_interpreter();
return (status);
}
ACPI_EXPORT_SYMBOL(acpi_acquire_global_lock)
/*******************************************************************************
*
* FUNCTION: acpi_release_global_lock
*
* PARAMETERS: handle - Returned from acpi_acquire_global_lock
*
* RETURN: Status
*
ACPICA 20050408 from Bob Moore Fixed three cases in the interpreter where an "index" argument to an ASL function was still (internally) 32 bits instead of the required 64 bits. This was the Index argument to the Index, Mid, and Match operators. The "strupr" function is now permanently local (acpi_ut_strupr), since this is not a POSIX-defined function and not present in most kernel-level C libraries. References to the C library strupr function have been removed from the headers. Completed the deployment of static functions/prototypes. All prototypes with the static attribute have been moved from the headers to the owning C file. ACPICA 20050329 from Bob Moore An error is now generated if an attempt is made to create a Buffer Field of length zero (A CreateField with a length operand of zero.) The interpreter now issues a warning whenever executable code at the module level is detected during ACPI table load. This will give some idea of the prevalence of this type of code. Implemented support for references to named objects (other than control methods) within package objects. Enhanced package object output for the debug object. Package objects are now completely dumped, showing all elements. Enhanced miscellaneous object output for the debug object. Any object can now be written to the debug object (for example, a device object can be written, and the type of the object will be displayed.) The "static" qualifier has been added to all local functions across the core subsystem. The number of "long" lines (> 80 chars) within the source has been significantly reduced, by about 1/3. Cleaned up all header files to ensure that all CA/iASL functions are prototyped (even static functions) and the formatting is consistent. Two new header files have been added, acopcode.h and acnames.h. Removed several obsolete functions that were no longer used. Signed-off-by: Len Brown <len.brown@intel.com>
2005-04-19 09:49:35 +07:00
* DESCRIPTION: Release the ACPI Global Lock. The handle must be valid.
*
******************************************************************************/
acpi_status acpi_release_global_lock(u32 handle)
{
acpi_status status;
if (!handle || (handle != acpi_gbl_global_lock_handle)) {
return (AE_NOT_ACQUIRED);
}
status = acpi_ex_release_mutex_object(acpi_gbl_global_lock_mutex);
return (status);
}
ACPI_EXPORT_SYMBOL(acpi_release_global_lock)
#endif /* !ACPI_REDUCED_HARDWARE */