aboutsummaryrefslogtreecommitdiff
path: root/drivers/gpu
diff options
context:
space:
mode:
authorSean Paul2019-09-18 16:07:29 -0400
committerSean Paul2019-09-19 10:03:32 -0400
commitd4da4e33341c5e6159543acc03559cb24f520bc2 (patch)
tree4fe613cff902325c07fd4e9706d7dcfc5035f115 /drivers/gpu
parent2d2e0b90a08f1e3a47b3ee852b27c219295423ef (diff)
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)
Diffstat (limited to 'drivers/gpu')
-rw-r--r--drivers/gpu/drm/drm_atomic_helper.c20
-rw-r--r--drivers/gpu/drm/drm_self_refresh_helper.c72
-rw-r--r--drivers/gpu/drm/rockchip/rockchip_drm_vop.c5
3 files changed, 87 insertions, 10 deletions
diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c
index 3a67f63c3146..3ef2ac52ce94 100644
--- a/drivers/gpu/drm/drm_atomic_helper.c
+++ b/drivers/gpu/drm/drm_atomic_helper.c
@@ -26,6 +26,7 @@
*/
#include <linux/dma-fence.h>
+#include <linux/ktime.h>
#include <drm/drm_atomic.h>
#include <drm/drm_atomic_helper.h>
@@ -1580,9 +1581,23 @@ static void commit_tail(struct drm_atomic_state *old_state)
{
struct drm_device *dev = old_state->dev;
const struct drm_mode_config_helper_funcs *funcs;
+ ktime_t start;
+ s64 commit_time_ms;
funcs = dev->mode_config.helper_private;
+ /*
+ * We're measuring the _entire_ commit, so the time will vary depending
+ * on how many fences and objects are involved. For the purposes of self
+ * refresh, this is desirable since it'll give us an idea of how
+ * congested things are. This will inform our decision on how often we
+ * should enter self refresh after idle.
+ *
+ * These times will be averaged out in the self refresh helpers to avoid
+ * overreacting over one outlier frame
+ */
+ start = ktime_get();
+
drm_atomic_helper_wait_for_fences(dev, old_state, false);
drm_atomic_helper_wait_for_dependencies(old_state);
@@ -1592,6 +1607,11 @@ static void commit_tail(struct drm_atomic_state *old_state)
else
drm_atomic_helper_commit_tail(old_state);
+ commit_time_ms = ktime_ms_delta(ktime_get(), start);
+ if (commit_time_ms > 0)
+ drm_self_refresh_helper_update_avg_times(old_state,
+ (unsigned long)commit_time_ms);
+
drm_atomic_helper_commit_cleanup_done(old_state);
drm_atomic_state_put(old_state);
diff --git a/drivers/gpu/drm/drm_self_refresh_helper.c b/drivers/gpu/drm/drm_self_refresh_helper.c
index 9095cebf2147..68f4765a5896 100644
--- a/drivers/gpu/drm/drm_self_refresh_helper.c
+++ b/drivers/gpu/drm/drm_self_refresh_helper.c
@@ -5,6 +5,7 @@
* Authors:
* Sean Paul <seanpaul@chromium.org>
*/
+#include <linux/average.h>
#include <linux/bitops.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
@@ -50,10 +51,17 @@
* 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;
- unsigned int entry_delay_ms;
+
+ 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)
@@ -122,6 +130,44 @@ out_drop_locks:
}
/**
+ * 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
*
@@ -152,6 +198,7 @@ void drm_self_refresh_helper_alter_state(struct drm_atomic_state *state)
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)
@@ -161,8 +208,13 @@ void drm_self_refresh_helper_alter_state(struct drm_atomic_state *state)
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(sr_data->entry_delay_ms));
+ msecs_to_jiffies(delay));
}
}
EXPORT_SYMBOL(drm_self_refresh_helper_alter_state);
@@ -170,12 +222,10 @@ 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
- * @entry_delay_ms: amount of inactivity to wait before entering self refresh
*
* Returns zero if successful or -errno on failure
*/
-int drm_self_refresh_helper_init(struct drm_crtc *crtc,
- unsigned int entry_delay_ms)
+int drm_self_refresh_helper_init(struct drm_crtc *crtc)
{
struct drm_self_refresh_data *sr_data = crtc->self_refresh_data;
@@ -189,8 +239,18 @@ int drm_self_refresh_helper_init(struct drm_crtc *crtc,
INIT_DELAYED_WORK(&sr_data->entry_work,
drm_self_refresh_helper_entry_work);
- sr_data->entry_delay_ms = entry_delay_ms;
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;
diff --git a/drivers/gpu/drm/rockchip/rockchip_drm_vop.c b/drivers/gpu/drm/rockchip/rockchip_drm_vop.c
index 2f821c58007c..613404f86668 100644
--- a/drivers/gpu/drm/rockchip/rockchip_drm_vop.c
+++ b/drivers/gpu/drm/rockchip/rockchip_drm_vop.c
@@ -39,8 +39,6 @@
#include "rockchip_drm_vop.h"
#include "rockchip_rgb.h"
-#define VOP_SELF_REFRESH_ENTRY_DELAY_MS 100
-
#define VOP_WIN_SET(vop, win, name, v) \
vop_reg_set(vop, &win->phy->name, win->base, ~0, v, #name)
#define VOP_SCL_SET(vop, win, name, v) \
@@ -1563,8 +1561,7 @@ static int vop_create_crtc(struct vop *vop)
init_completion(&vop->line_flag_completion);
crtc->port = port;
- ret = drm_self_refresh_helper_init(crtc,
- VOP_SELF_REFRESH_ENTRY_DELAY_MS);
+ ret = drm_self_refresh_helper_init(crtc);
if (ret)
DRM_DEV_DEBUG_KMS(vop->dev,
"Failed to init %s with SR helpers %d, ignoring\n",