/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
/*
*
* (C) COPYRIGHT 2019-2022 ARM Limited. All rights reserved.
*
* This program is free software and is provided to you under the terms of the
* GNU General Public License version 2 as published by the Free Software
* Foundation, and any use by you of this program is subject to the terms
* of such GNU license.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, you can access it online at
* http://www.gnu.org/licenses/gpl-2.0.html.
*
*/
#ifndef _KBASE_CSF_SCHEDULER_H_
#define _KBASE_CSF_SCHEDULER_H_
#include "mali_kbase_csf.h"
#include "mali_kbase_csf_event.h"
/**
* kbase_csf_scheduler_queue_start() - Enable the running of GPU command queue
* on firmware.
*
* @queue: Pointer to the GPU command queue to be started.
*
* This function would enable the start of a CSI, within a
* CSG, to which the @queue was bound.
* If the CSG is already scheduled and resident, the CSI will be started
* right away, otherwise once the group is made resident.
*
* Return: 0 on success, or negative on failure.
*/
int kbase_csf_scheduler_queue_start(struct kbase_queue *queue);
/**
* kbase_csf_scheduler_queue_stop() - Disable the running of GPU command queue
* on firmware.
*
* @queue: Pointer to the GPU command queue to be stopped.
*
* This function would stop the CSI, within a CSG, to which @queue was bound.
*
* Return: 0 on success, or negative on failure.
*/
int kbase_csf_scheduler_queue_stop(struct kbase_queue *queue);
/**
* kbase_csf_scheduler_group_protm_enter - Handle the protm enter event for the
* GPU command queue group.
*
* @group: The command queue group.
*
* This function could request the firmware to enter the protected mode
* and allow the execution of protected region instructions for all the
* bound queues of the group that have protm pending bit set in their
* respective CS_ACK register.
*/
void kbase_csf_scheduler_group_protm_enter(struct kbase_queue_group *group);
/**
* kbase_csf_scheduler_group_get_slot() - Checks if a queue group is
* programmed on a firmware CSG slot
* and returns the slot number.
*
* @group: The command queue group.
*
* Return: The slot number, if the group is programmed on a slot.
* Otherwise returns a negative number.
*
* Note: This function should not be used if the interrupt_lock is held. Use
* kbase_csf_scheduler_group_get_slot_locked() instead.
*/
int kbase_csf_scheduler_group_get_slot(struct kbase_queue_group *group);
/**
* kbase_csf_scheduler_group_get_slot_locked() - Checks if a queue group is
* programmed on a firmware CSG slot
* and returns the slot number.
*
* @group: The command queue group.
*
* Return: The slot number, if the group is programmed on a slot.
* Otherwise returns a negative number.
*
* Note: Caller must hold the interrupt_lock.
*/
int kbase_csf_scheduler_group_get_slot_locked(struct kbase_queue_group *group);
/**
* kbase_csf_scheduler_group_events_enabled() - Checks if interrupt events
* should be handled for a queue group.
*
* @kbdev: The device of the group.
* @group: The queue group.
*
* Return: true if interrupt events should be handled.
*
* Note: Caller must hold the interrupt_lock.
*/
bool kbase_csf_scheduler_group_events_enabled(struct kbase_device *kbdev,
struct kbase_queue_group *group);
/**
* kbase_csf_scheduler_get_group_on_slot()- Gets the queue group that has been
* programmed to a firmware CSG slot.
*
* @kbdev: The GPU device.
* @slot: The slot for which to get the queue group.
*
* Return: Pointer to the programmed queue group.
*
* Note: Caller must hold the interrupt_lock.
*/
struct kbase_queue_group *kbase_csf_scheduler_get_group_on_slot(
struct kbase_device *kbdev, int slot);
/**
* kbase_csf_scheduler_group_deschedule() - Deschedule a GPU command queue
* group from the firmware.
*
* @group: Pointer to the queue group to be descheduled.
*
* This function would disable the scheduling of GPU command queue group on
* firmware.
*/
void kbase_csf_scheduler_group_deschedule(struct kbase_queue_group *group);
/**
* kbase_csf_scheduler_evict_ctx_slots() - Evict all GPU command queue groups
* of a given context that are active
* running from the firmware.
*
* @kbdev: The GPU device.
* @kctx: Kbase context for the evict operation.
* @evicted_groups: List_head for returning evicted active queue groups.
*
* This function would disable the scheduling of GPU command queue groups active
* on firmware slots from the given Kbase context. The affected groups are
* added to the supplied list_head argument.
*/
void kbase_csf_scheduler_evict_ctx_slots(struct kbase_device *kbdev,
struct kbase_context *kctx, struct list_head *evicted_groups);
/**
* kbase_csf_scheduler_context_init() - Initialize the context-specific part
* for CSF scheduler.
*
* @kctx: Pointer to kbase context that is being created.
*
* This function must be called during Kbase context creation.
*
* Return: 0 on success, or negative on failure.
*/
int kbase_csf_scheduler_context_init(struct kbase_context *kctx);
/**
* kbase_csf_scheduler_init - Initialize the CSF scheduler
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* The scheduler does the arbitration for the CSG slots
* provided by the firmware between the GPU command queue groups created
* by the Clients.
* This function must be called after loading firmware and parsing its capabilities.
*
* Return: 0 on success, or negative on failure.
*/
int kbase_csf_scheduler_init(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_early_init - Early initialization for the CSF scheduler
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* Initialize necessary resources such as locks, workqueue for CSF scheduler.
* This must be called at kbase probe.
*
* Return: 0 on success, or negative on failure.
*/
int kbase_csf_scheduler_early_init(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_context_term() - Terminate the context-specific part
* for CSF scheduler.
*
* @kctx: Pointer to kbase context that is being terminated.
*
* This function must be called during Kbase context termination.
*/
void kbase_csf_scheduler_context_term(struct kbase_context *kctx);
/**
* kbase_csf_scheduler_term - Terminate the CSF scheduler.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This should be called when unload of firmware is done on device
* termination.
*/
void kbase_csf_scheduler_term(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_early_term - Early termination of the CSF scheduler.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This should be called only when kbase probe fails or gets rmmoded.
*/
void kbase_csf_scheduler_early_term(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_reset - Reset the state of all active GPU command
* queue groups.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will first iterate through all the active/scheduled GPU
* command queue groups and suspend them (to avoid losing work for groups
* that are not stuck). The groups that could not get suspended would be
* descheduled and marked as terminated (which will then lead to unbinding
* of all the queues bound to them) and also no more work would be allowed
* to execute for them.
*
* This is similar to the action taken in response to an unexpected OoM event.
* No explicit re-initialization is done for CSG & CS interface I/O pages;
* instead, that happens implicitly on firmware reload.
*
* Should be called only after initiating the GPU reset.
*/
void kbase_csf_scheduler_reset(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_enable_tick_timer - Enable the scheduler tick timer.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will restart the scheduler tick so that regular scheduling can
* be resumed without any explicit trigger (like kicking of GPU queues).
*/
void kbase_csf_scheduler_enable_tick_timer(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_group_copy_suspend_buf - Suspend a queue
* group and copy suspend buffer.
*
* @group: Pointer to the queue group to be suspended.
* @sus_buf: Pointer to the structure which contains details of the
* user buffer and its kernel pinned pages to which we need to copy
* the group suspend buffer.
*
* This function is called to suspend a queue group and copy the suspend_buffer
* contents to the input buffer provided.
*
* Return: 0 on success, or negative on failure.
*/
int kbase_csf_scheduler_group_copy_suspend_buf(struct kbase_queue_group *group,
struct kbase_suspend_copy_buffer *sus_buf);
/**
* kbase_csf_scheduler_lock - Acquire the global Scheduler lock.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will take the global scheduler lock, in order to serialize
* against the Scheduler actions, for access to CS IO pages.
*/
static inline void kbase_csf_scheduler_lock(struct kbase_device *kbdev)
{
mutex_lock(&kbdev->csf.scheduler.lock);
}
/**
* kbase_csf_scheduler_unlock - Release the global Scheduler lock.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*/
static inline void kbase_csf_scheduler_unlock(struct kbase_device *kbdev)
{
mutex_unlock(&kbdev->csf.scheduler.lock);
}
/**
* kbase_csf_scheduler_spin_lock - Acquire Scheduler interrupt spinlock.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
* @flags: Pointer to the memory location that would store the previous
* interrupt state.
*
* This function will take the global scheduler lock, in order to serialize
* against the Scheduler actions, for access to CS IO pages.
*/
static inline void kbase_csf_scheduler_spin_lock(struct kbase_device *kbdev,
unsigned long *flags)
{
spin_lock_irqsave(&kbdev->csf.scheduler.interrupt_lock, *flags);
}
/**
* kbase_csf_scheduler_spin_unlock - Release Scheduler interrupt spinlock.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
* @flags: Previously stored interrupt state when Scheduler interrupt
* spinlock was acquired.
*/
static inline void kbase_csf_scheduler_spin_unlock(struct kbase_device *kbdev,
unsigned long flags)
{
spin_unlock_irqrestore(&kbdev->csf.scheduler.interrupt_lock, flags);
}
/**
* kbase_csf_scheduler_spin_lock_assert_held - Assert if the Scheduler
* interrupt spinlock is held.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*/
static inline void
kbase_csf_scheduler_spin_lock_assert_held(struct kbase_device *kbdev)
{
lockdep_assert_held(&kbdev->csf.scheduler.interrupt_lock);
}
/**
* kbase_csf_scheduler_timer_is_enabled() - Check if the scheduler wakes up
* automatically for periodic tasks.
*
* @kbdev: Pointer to the device
*
* Return: true if the scheduler is configured to wake up periodically
*/
bool kbase_csf_scheduler_timer_is_enabled(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_timer_set_enabled() - Enable/disable periodic
* scheduler tasks.
*
* @kbdev: Pointer to the device
* @enable: Whether to enable periodic scheduler tasks
*/
void kbase_csf_scheduler_timer_set_enabled(struct kbase_device *kbdev,
bool enable);
/**
* kbase_csf_scheduler_kick - Perform pending scheduling tasks once.
*
* Note: This function is only effective if the scheduling timer is disabled.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*/
void kbase_csf_scheduler_kick(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_protected_mode_in_use() - Check if the scheduler is
* running with protected mode tasks.
*
* @kbdev: Pointer to the device
*
* Return: true if the scheduler is running with protected mode tasks
*/
static inline bool kbase_csf_scheduler_protected_mode_in_use(
struct kbase_device *kbdev)
{
return (kbdev->csf.scheduler.active_protm_grp != NULL);
}
/**
* kbase_csf_scheduler_pm_active - Perform scheduler power active operation
*
* Note: This function will increase the scheduler's internal pm_active_count
* value, ensuring that both GPU and MCU are powered for access. The MCU may
* not have actually become active when this function returns, so need to
* call kbase_csf_scheduler_wait_mcu_active() for that.
*
* This function should not be called with global scheduler lock held.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*/
void kbase_csf_scheduler_pm_active(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_pm_idle - Perform the scheduler power idle operation
*
* Note: This function will decrease the scheduler's internal pm_active_count
* value. On reaching 0, the MCU and GPU could be powered off. This function
* should not be called with global scheduler lock held.
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*/
void kbase_csf_scheduler_pm_idle(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_wait_mcu_active - Wait for the MCU to actually become active
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will wait for the MCU to actually become active. It is supposed
* to be called after calling kbase_csf_scheduler_pm_active(). It is needed as
* kbase_csf_scheduler_pm_active() may not make the MCU active right away.
*
* Return: 0 if the MCU was successfully activated otherwise an error code.
*/
int kbase_csf_scheduler_wait_mcu_active(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_pm_resume_no_lock - Reactivate the scheduler on system resume
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will make the scheduler resume the scheduling of queue groups
* and take the power managemenet reference, if there are any runnable groups.
* The caller must have acquired the global Scheduler lock.
*/
void kbase_csf_scheduler_pm_resume_no_lock(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_pm_resume - Reactivate the scheduler on system resume
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will make the scheduler resume the scheduling of queue groups
* and take the power managemenet reference, if there are any runnable groups.
*/
void kbase_csf_scheduler_pm_resume(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_pm_suspend_no_lock - Idle the scheduler on system suspend
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will make the scheduler suspend all the running queue groups
* and drop its power managemenet reference.
* The caller must have acquired the global Scheduler lock.
*
* Return: 0 on success.
*/
int kbase_csf_scheduler_pm_suspend_no_lock(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_pm_suspend - Idle the scheduler on system suspend
*
* @kbdev: Instance of a GPU platform device that implements a CSF interface.
*
* This function will make the scheduler suspend all the running queue groups
* and drop its power managemenet reference.
*
* Return: 0 on success.
*/
int kbase_csf_scheduler_pm_suspend(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_all_csgs_idle() - Check if the scheduler internal
* runtime used slots are all tagged as idle command queue groups.
*
* @kbdev: Pointer to the device
*
* Return: true if all the used slots are tagged as idle CSGs.
*/
static inline bool kbase_csf_scheduler_all_csgs_idle(struct kbase_device *kbdev)
{
lockdep_assert_held(&kbdev->csf.scheduler.interrupt_lock);
return bitmap_equal(kbdev->csf.scheduler.csg_slots_idle_mask,
kbdev->csf.scheduler.csg_inuse_bitmap,
kbdev->csf.global_iface.group_num);
}
/**
* kbase_csf_scheduler_tick_advance_nolock() - Advance the scheduling tick
*
* @kbdev: Pointer to the device
*
* This function advances the scheduling tick by enqueing the tick work item for
* immediate execution, but only if the tick hrtimer is active. If the timer
* is inactive then the tick work item is already in flight.
* The caller must hold the interrupt lock.
*/
static inline void
kbase_csf_scheduler_tick_advance_nolock(struct kbase_device *kbdev)
{
struct kbase_csf_scheduler *const scheduler = &kbdev->csf.scheduler;
lockdep_assert_held(&scheduler->interrupt_lock);
if (scheduler->tick_timer_active) {
KBASE_KTRACE_ADD(kbdev, SCHEDULER_TICK_ADVANCE, NULL, 0u);
scheduler->tick_timer_active = false;
queue_work(scheduler->wq, &scheduler->tick_work);
} else {
KBASE_KTRACE_ADD(kbdev, SCHEDULER_TICK_NOADVANCE, NULL, 0u);
}
}
/**
* kbase_csf_scheduler_tick_advance() - Advance the scheduling tick
*
* @kbdev: Pointer to the device
*
* This function advances the scheduling tick by enqueing the tick work item for
* immediate execution, but only if the tick hrtimer is active. If the timer
* is inactive then the tick work item is already in flight.
*/
static inline void kbase_csf_scheduler_tick_advance(struct kbase_device *kbdev)
{
struct kbase_csf_scheduler *const scheduler = &kbdev->csf.scheduler;
unsigned long flags;
spin_lock_irqsave(&scheduler->interrupt_lock, flags);
kbase_csf_scheduler_tick_advance_nolock(kbdev);
spin_unlock_irqrestore(&scheduler->interrupt_lock, flags);
}
/**
* kbase_csf_scheduler_invoke_tick() - Invoke the scheduling tick
*
* @kbdev: Pointer to the device
*
* This function will queue the scheduling tick work item for immediate
* execution if tick timer is not active. This can be called from interrupt
* context to resume the scheduling after GPU was put to sleep.
*/
static inline void kbase_csf_scheduler_invoke_tick(struct kbase_device *kbdev)
{
struct kbase_csf_scheduler *const scheduler = &kbdev->csf.scheduler;
unsigned long flags;
spin_lock_irqsave(&scheduler->interrupt_lock, flags);
if (!scheduler->tick_timer_active)
queue_work(scheduler->wq, &scheduler->tick_work);
spin_unlock_irqrestore(&scheduler->interrupt_lock, flags);
}
/**
* kbase_csf_scheduler_queue_has_trace() - report whether the queue has been
* configured to operate with the
* cs_trace feature.
*
* @queue: Pointer to the queue.
*
* Return: True if the gpu queue is configured to operate with the cs_trace
* feature, otherwise false.
*/
static inline bool kbase_csf_scheduler_queue_has_trace(struct kbase_queue *queue)
{
lockdep_assert_held(&queue->kctx->kbdev->csf.scheduler.lock);
/* In the current arrangement, it is possible for the context to enable
* the cs_trace after some queues have been registered with cs_trace in
* disabled state. So each queue has its own enabled/disabled condition.
*/
return (queue->trace_buffer_size && queue->trace_buffer_base);
}
#ifdef KBASE_PM_RUNTIME
/**
* kbase_csf_scheduler_reval_idleness_post_sleep() - Check GPU's idleness after
* putting MCU to sleep state
*
* @kbdev: Pointer to the device
*
* This function re-evaluates the idleness of on-slot queue groups after MCU
* was put to the sleep state and invokes the scheduling tick if any of the
* on-slot queue group became non-idle.
* CSG_OUTPUT_BLOCK.CSG_STATUS_STATE.IDLE bit is checked to determine the
* idleness which is updated by MCU firmware on handling of the sleep request.
*
* This function is needed to detect if more work was flushed in the window
* between the GPU idle notification and the enabling of Doorbell mirror
* interrupt (from MCU state machine). Once Doorbell mirror interrupt is
* enabled, Host can receive the notification on User doorbell rings.
*/
void kbase_csf_scheduler_reval_idleness_post_sleep(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_handle_runtime_suspend() - Handle runtime suspend by
* suspending CSGs.
*
* @kbdev: Pointer to the device
*
* This function is called from the runtime suspend callback function for
* suspending all the on-slot queue groups. If any of the group is found to
* be non-idle after the completion of CSG suspend operation or the CSG
* suspend operation times out, then the scheduling tick is invoked and an
* error is returned so that the GPU power down can be aborted.
*
* Return: 0 if all the CSGs were suspended, otherwise an error code.
*/
int kbase_csf_scheduler_handle_runtime_suspend(struct kbase_device *kbdev);
#endif
/**
* kbase_csf_scheduler_process_gpu_idle_event() - Process GPU idle IRQ
*
* @kbdev: Pointer to the device
*
* This function is called when a GPU idle IRQ has been raised.
*/
void kbase_csf_scheduler_process_gpu_idle_event(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_get_nr_active_csgs() - Get the number of active CSGs
*
* @kbdev: Pointer to the device
*
* This function calculates the number of CSG slots that have a queue group
* resident on them.
*
* Note: This function should not be used if the interrupt_lock is held. Use
* kbase_csf_scheduler_get_nr_active_csgs_locked() instead.
*
* Return: number of active CSGs.
*/
u32 kbase_csf_scheduler_get_nr_active_csgs(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_get_nr_active_csgs_locked() - Get the number of active
* CSGs
*
* @kbdev: Pointer to the device
*
* This function calculates the number of CSG slots that have a queue group
* resident on them.
*
* Note: This function should be called with interrupt_lock held.
*
* Return: number of active CSGs.
*/
u32 kbase_csf_scheduler_get_nr_active_csgs_locked(struct kbase_device *kbdev);
/**
* kbase_csf_scheduler_force_wakeup() - Forcefully resume the scheduling of CSGs
*
* @kbdev: Pointer to the device
*
* This function is called to forcefully resume the scheduling of CSGs, even
* when there wasn't any work submitted for them.
* This function is only used for testing purpose.
*/
void kbase_csf_scheduler_force_wakeup(struct kbase_device *kbdev);
#ifdef KBASE_PM_RUNTIME
/**
* kbase_csf_scheduler_force_sleep() - Forcefully put the Scheduler to sleeping
* state.
*
* @kbdev: Pointer to the device
*
* This function is called to forcefully put the Scheduler to sleeping state
* and trigger the sleep of MCU. If the CSGs are not idle, then the Scheduler
* would get reactivated again immediately.
* This function is only used for testing purpose.
*/
void kbase_csf_scheduler_force_sleep(struct kbase_device *kbdev);
#endif
#endif /* _KBASE_CSF_SCHEDULER_H_ */
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags