linux/drivers/gpu/drm/drm_self_refresh_helper.c
Sean Paul d4da4e3334 drm: Measure Self Refresh Entry/Exit times to avoid thrashing
Currently the self refresh idle timer is a const set by the crtc. This
is fine if the self refresh entry/exit times are well-known for all
panels used on that crtc. However panels and workloads can vary quite a
bit, and a timeout which works well for one doesn't work well for
another.

In the extreme, if the timeout is too short we could get in a situation
where the self refresh exits are taking so long we queue up a self refresh
entry before the exit commit is even finished.

This patch changes the idle timeout to a moving average of the entry
times + a moving average of exit times + the crtc constant.

This patch was tested on rockchip, with a kevin CrOS panel the idle
delay averages out to about ~235ms (35 entry + 100 exit + 100 const). On
the same board, the bob panel idle delay lands around ~340ms (90 entry
+ 150 exit + 100 const).

WRT the dedicated mutex in self_refresh_data, it would be nice if we
could rely on drm_crtc.mutex to protect the average times, but there are
a few reasons why a separate lock is a better choice:
- We can't rely on drm_crtc.mutex being held if we're doing a nonblocking
  commit
- We can't grab drm_crtc.mutex since drm_modeset_lock() doesn't tell us
  whether the lock was already held in the acquire context (it eats
  -EALREADY), so we can't tell if we should drop it or not
- We don't need such a heavy-handed lock for what we're trying to do,
  commit ordering doesn't matter, so a point-of-use lock will be less
  contentious

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link to v1: https://patchwork.freedesktop.org/patch/msgid/20190917200443.64481-2-sean@poorly.run
Link: https://patchwork.freedesktop.org/patch/msgid/20190918200734.149876-2-sean@poorly.run

Changes in v2:
- Migrate locking explanation from comment to commit msg (Daniel)
- Turf constant entry delay and multiply the avg times by 2 (Daniel)
2019-09-19 10:03:32 -04:00

278 lines
8.1 KiB
C

// SPDX-License-Identifier: MIT
/*
* Copyright (C) 2019 Google, Inc.
*
* Authors:
* Sean Paul <seanpaul@chromium.org>
*/
#include <linux/average.h>
#include <linux/bitops.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include <drm/drm_atomic.h>
#include <drm/drm_atomic_helper.h>
#include <drm/drm_connector.h>
#include <drm/drm_crtc.h>
#include <drm/drm_device.h>
#include <drm/drm_mode_config.h>
#include <drm/drm_modeset_lock.h>
#include <drm/drm_print.h>
#include <drm/drm_self_refresh_helper.h>
/**
* DOC: overview
*
* This helper library provides an easy way for drivers to leverage the atomic
* framework to implement panel self refresh (SR) support. Drivers are
* responsible for initializing and cleaning up the SR helpers on load/unload
* (see &drm_self_refresh_helper_init/&drm_self_refresh_helper_cleanup).
* The connector is responsible for setting
* &drm_connector_state.self_refresh_aware to true at runtime if it is SR-aware
* (meaning it knows how to initiate self refresh on the panel).
*
* Once a crtc has enabled SR using &drm_self_refresh_helper_init, the
* helpers will monitor activity and call back into the driver to enable/disable
* SR as appropriate. The best way to think about this is that it's a DPMS
* on/off request with &drm_crtc_state.self_refresh_active set in crtc state
* that tells you to disable/enable SR on the panel instead of power-cycling it.
*
* During SR, drivers may choose to fully disable their crtc/encoder/bridge
* hardware (in which case no driver changes are necessary), or they can inspect
* &drm_crtc_state.self_refresh_active if they want to enter low power mode
* without full disable (in case full disable/enable is too slow).
*
* SR will be deactivated if there are any atomic updates affecting the
* pipe that is in SR mode. If a crtc is driving multiple connectors, all
* connectors must be SR aware and all will enter/exit SR mode at the same time.
*
* If the crtc and connector are SR aware, but the panel connected does not
* support it (or is otherwise unable to enter SR), the driver should fail
* atomic_check when &drm_crtc_state.self_refresh_active is true.
*/
#define SELF_REFRESH_AVG_SEED_MS 200
DECLARE_EWMA(psr_time, 4, 4)
struct drm_self_refresh_data {
struct drm_crtc *crtc;
struct delayed_work entry_work;
struct mutex avg_mutex;
struct ewma_psr_time entry_avg_ms;
struct ewma_psr_time exit_avg_ms;
};
static void drm_self_refresh_helper_entry_work(struct work_struct *work)
{
struct drm_self_refresh_data *sr_data = container_of(
to_delayed_work(work),
struct drm_self_refresh_data, entry_work);
struct drm_crtc *crtc = sr_data->crtc;
struct drm_device *dev = crtc->dev;
struct drm_modeset_acquire_ctx ctx;
struct drm_atomic_state *state;
struct drm_connector *conn;
struct drm_connector_state *conn_state;
struct drm_crtc_state *crtc_state;
int i, ret = 0;
drm_modeset_acquire_init(&ctx, 0);
state = drm_atomic_state_alloc(dev);
if (!state) {
ret = -ENOMEM;
goto out_drop_locks;
}
retry:
state->acquire_ctx = &ctx;
crtc_state = drm_atomic_get_crtc_state(state, crtc);
if (IS_ERR(crtc_state)) {
ret = PTR_ERR(crtc_state);
goto out;
}
if (!crtc_state->enable)
goto out;
ret = drm_atomic_add_affected_connectors(state, crtc);
if (ret)
goto out;
for_each_new_connector_in_state(state, conn, conn_state, i) {
if (!conn_state->self_refresh_aware)
goto out;
}
crtc_state->active = false;
crtc_state->self_refresh_active = true;
ret = drm_atomic_commit(state);
if (ret)
goto out;
out:
if (ret == -EDEADLK) {
drm_atomic_state_clear(state);
ret = drm_modeset_backoff(&ctx);
if (!ret)
goto retry;
}
drm_atomic_state_put(state);
out_drop_locks:
drm_modeset_drop_locks(&ctx);
drm_modeset_acquire_fini(&ctx);
}
/**
* drm_self_refresh_helper_update_avg_times - Updates a crtc's SR time averages
* @state: the state which has just been applied to hardware
* @commit_time_ms: the amount of time in ms that this commit took to complete
*
* Called after &drm_mode_config_funcs.atomic_commit_tail, this function will
* update the average entry/exit self refresh times on self refresh transitions.
* These averages will be used when calculating how long to delay before
* entering self refresh mode after activity.
*/
void drm_self_refresh_helper_update_avg_times(struct drm_atomic_state *state,
unsigned int commit_time_ms)
{
struct drm_crtc *crtc;
struct drm_crtc_state *old_crtc_state, *new_crtc_state;
int i;
for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state,
new_crtc_state, i) {
struct drm_self_refresh_data *sr_data = crtc->self_refresh_data;
struct ewma_psr_time *time;
if (old_crtc_state->self_refresh_active ==
new_crtc_state->self_refresh_active)
continue;
if (new_crtc_state->self_refresh_active)
time = &sr_data->entry_avg_ms;
else
time = &sr_data->exit_avg_ms;
mutex_lock(&sr_data->avg_mutex);
ewma_psr_time_add(time, commit_time_ms);
mutex_unlock(&sr_data->avg_mutex);
}
}
EXPORT_SYMBOL(drm_self_refresh_helper_update_avg_times);
/**
* drm_self_refresh_helper_alter_state - Alters the atomic state for SR exit
* @state: the state currently being checked
*
* Called at the end of atomic check. This function checks the state for flags
* incompatible with self refresh exit and changes them. This is a bit
* disingenuous since userspace is expecting one thing and we're giving it
* another. However in order to keep self refresh entirely hidden from
* userspace, this is required.
*
* At the end, we queue up the self refresh entry work so we can enter PSR after
* the desired delay.
*/
void drm_self_refresh_helper_alter_state(struct drm_atomic_state *state)
{
struct drm_crtc *crtc;
struct drm_crtc_state *crtc_state;
int i;
if (state->async_update || !state->allow_modeset) {
for_each_old_crtc_in_state(state, crtc, crtc_state, i) {
if (crtc_state->self_refresh_active) {
state->async_update = false;
state->allow_modeset = true;
break;
}
}
}
for_each_new_crtc_in_state(state, crtc, crtc_state, i) {
struct drm_self_refresh_data *sr_data;
unsigned int delay;
/* Don't trigger the entry timer when we're already in SR */
if (crtc_state->self_refresh_active)
continue;
sr_data = crtc->self_refresh_data;
if (!sr_data)
continue;
mutex_lock(&sr_data->avg_mutex);
delay = (ewma_psr_time_read(&sr_data->entry_avg_ms) +
ewma_psr_time_read(&sr_data->exit_avg_ms)) * 2;
mutex_unlock(&sr_data->avg_mutex);
mod_delayed_work(system_wq, &sr_data->entry_work,
msecs_to_jiffies(delay));
}
}
EXPORT_SYMBOL(drm_self_refresh_helper_alter_state);
/**
* drm_self_refresh_helper_init - Initializes self refresh helpers for a crtc
* @crtc: the crtc which supports self refresh supported displays
*
* Returns zero if successful or -errno on failure
*/
int drm_self_refresh_helper_init(struct drm_crtc *crtc)
{
struct drm_self_refresh_data *sr_data = crtc->self_refresh_data;
/* Helper is already initialized */
if (WARN_ON(sr_data))
return -EINVAL;
sr_data = kzalloc(sizeof(*sr_data), GFP_KERNEL);
if (!sr_data)
return -ENOMEM;
INIT_DELAYED_WORK(&sr_data->entry_work,
drm_self_refresh_helper_entry_work);
sr_data->crtc = crtc;
mutex_init(&sr_data->avg_mutex);
ewma_psr_time_init(&sr_data->entry_avg_ms);
ewma_psr_time_init(&sr_data->exit_avg_ms);
/*
* Seed the averages so they're non-zero (and sufficiently large
* for even poorly performing panels). As time goes on, this will be
* averaged out and the values will trend to their true value.
*/
ewma_psr_time_add(&sr_data->entry_avg_ms, SELF_REFRESH_AVG_SEED_MS);
ewma_psr_time_add(&sr_data->exit_avg_ms, SELF_REFRESH_AVG_SEED_MS);
crtc->self_refresh_data = sr_data;
return 0;
}
EXPORT_SYMBOL(drm_self_refresh_helper_init);
/**
* drm_self_refresh_helper_cleanup - Cleans up self refresh helpers for a crtc
* @crtc: the crtc to cleanup
*/
void drm_self_refresh_helper_cleanup(struct drm_crtc *crtc)
{
struct drm_self_refresh_data *sr_data = crtc->self_refresh_data;
/* Helper is already uninitialized */
if (!sr_data)
return;
crtc->self_refresh_data = NULL;
cancel_delayed_work_sync(&sr_data->entry_work);
kfree(sr_data);
}
EXPORT_SYMBOL(drm_self_refresh_helper_cleanup);