drm/i915/display: move plane prepare/cleanup to intel_atomic_plane.c

Start to refactor more stuff out of intel_display.c. These fit
better in this file.

This moves the rps boosting code as well as this is the only user of it.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211012043502.1377715-2-airlied@gmail.com
This commit is contained in:
Dave Airlie 2021-10-12 14:34:55 +10:00 committed by Jani Nikula
parent 5c31e9d013
commit 74a75dc908
3 changed files with 208 additions and 212 deletions

View File

@ -41,6 +41,7 @@
#include "intel_display_types.h"
#include "intel_pm.h"
#include "intel_sprite.h"
#include "gt/intel_rps.h"
static void intel_plane_state_reset(struct intel_plane_state *plane_state,
struct intel_plane *plane)
@ -601,6 +602,213 @@ int intel_atomic_plane_check_clipping(struct intel_plane_state *plane_state,
return 0;
}
struct wait_rps_boost {
struct wait_queue_entry wait;
struct drm_crtc *crtc;
struct i915_request *request;
};
static int do_rps_boost(struct wait_queue_entry *_wait,
unsigned mode, int sync, void *key)
{
struct wait_rps_boost *wait = container_of(_wait, typeof(*wait), wait);
struct i915_request *rq = wait->request;
/*
* If we missed the vblank, but the request is already running it
* is reasonable to assume that it will complete before the next
* vblank without our intervention, so leave RPS alone.
*/
if (!i915_request_started(rq))
intel_rps_boost(rq);
i915_request_put(rq);
drm_crtc_vblank_put(wait->crtc);
list_del(&wait->wait.entry);
kfree(wait);
return 1;
}
static void add_rps_boost_after_vblank(struct drm_crtc *crtc,
struct dma_fence *fence)
{
struct wait_rps_boost *wait;
if (!dma_fence_is_i915(fence))
return;
if (DISPLAY_VER(to_i915(crtc->dev)) < 6)
return;
if (drm_crtc_vblank_get(crtc))
return;
wait = kmalloc(sizeof(*wait), GFP_KERNEL);
if (!wait) {
drm_crtc_vblank_put(crtc);
return;
}
wait->request = to_request(dma_fence_get(fence));
wait->crtc = crtc;
wait->wait.func = do_rps_boost;
wait->wait.flags = 0;
add_wait_queue(drm_crtc_vblank_waitqueue(crtc), &wait->wait);
}
/**
* intel_prepare_plane_fb - Prepare fb for usage on plane
* @_plane: drm plane to prepare for
* @_new_plane_state: the plane state being prepared
*
* Prepares a framebuffer for usage on a display plane. Generally this
* involves pinning the underlying object and updating the frontbuffer tracking
* bits. Some older platforms need special physical address handling for
* cursor planes.
*
* Returns 0 on success, negative error code on failure.
*/
static int
intel_prepare_plane_fb(struct drm_plane *_plane,
struct drm_plane_state *_new_plane_state)
{
struct i915_sched_attr attr = { .priority = I915_PRIORITY_DISPLAY };
struct intel_plane *plane = to_intel_plane(_plane);
struct intel_plane_state *new_plane_state =
to_intel_plane_state(_new_plane_state);
struct intel_atomic_state *state =
to_intel_atomic_state(new_plane_state->uapi.state);
struct drm_i915_private *dev_priv = to_i915(plane->base.dev);
const struct intel_plane_state *old_plane_state =
intel_atomic_get_old_plane_state(state, plane);
struct drm_i915_gem_object *obj = intel_fb_obj(new_plane_state->hw.fb);
struct drm_i915_gem_object *old_obj = intel_fb_obj(old_plane_state->hw.fb);
int ret;
if (old_obj) {
const struct intel_crtc_state *crtc_state =
intel_atomic_get_new_crtc_state(state,
to_intel_crtc(old_plane_state->hw.crtc));
/* Big Hammer, we also need to ensure that any pending
* MI_WAIT_FOR_EVENT inside a user batch buffer on the
* current scanout is retired before unpinning the old
* framebuffer. Note that we rely on userspace rendering
* into the buffer attached to the pipe they are waiting
* on. If not, userspace generates a GPU hang with IPEHR
* point to the MI_WAIT_FOR_EVENT.
*
* This should only fail upon a hung GPU, in which case we
* can safely continue.
*/
if (intel_crtc_needs_modeset(crtc_state)) {
ret = i915_sw_fence_await_reservation(&state->commit_ready,
old_obj->base.resv, NULL,
false, 0,
GFP_KERNEL);
if (ret < 0)
return ret;
}
}
if (new_plane_state->uapi.fence) { /* explicit fencing */
i915_gem_fence_wait_priority(new_plane_state->uapi.fence,
&attr);
ret = i915_sw_fence_await_dma_fence(&state->commit_ready,
new_plane_state->uapi.fence,
i915_fence_timeout(dev_priv),
GFP_KERNEL);
if (ret < 0)
return ret;
}
if (!obj)
return 0;
ret = intel_plane_pin_fb(new_plane_state);
if (ret)
return ret;
i915_gem_object_wait_priority(obj, 0, &attr);
if (!new_plane_state->uapi.fence) { /* implicit fencing */
struct dma_fence *fence;
ret = i915_sw_fence_await_reservation(&state->commit_ready,
obj->base.resv, NULL,
false,
i915_fence_timeout(dev_priv),
GFP_KERNEL);
if (ret < 0)
goto unpin_fb;
fence = dma_resv_get_excl_unlocked(obj->base.resv);
if (fence) {
add_rps_boost_after_vblank(new_plane_state->hw.crtc,
fence);
dma_fence_put(fence);
}
} else {
add_rps_boost_after_vblank(new_plane_state->hw.crtc,
new_plane_state->uapi.fence);
}
/*
* We declare pageflips to be interactive and so merit a small bias
* towards upclocking to deliver the frame on time. By only changing
* the RPS thresholds to sample more regularly and aim for higher
* clocks we can hopefully deliver low power workloads (like kodi)
* that are not quite steady state without resorting to forcing
* maximum clocks following a vblank miss (see do_rps_boost()).
*/
if (!state->rps_interactive) {
intel_rps_mark_interactive(&dev_priv->gt.rps, true);
state->rps_interactive = true;
}
return 0;
unpin_fb:
intel_plane_unpin_fb(new_plane_state);
return ret;
}
/**
* intel_cleanup_plane_fb - Cleans up an fb after plane use
* @plane: drm plane to clean up for
* @_old_plane_state: the state from the previous modeset
*
* Cleans up a framebuffer that has just been removed from a plane.
*/
static void
intel_cleanup_plane_fb(struct drm_plane *plane,
struct drm_plane_state *_old_plane_state)
{
struct intel_plane_state *old_plane_state =
to_intel_plane_state(_old_plane_state);
struct intel_atomic_state *state =
to_intel_atomic_state(old_plane_state->uapi.state);
struct drm_i915_private *dev_priv = to_i915(plane->dev);
struct drm_i915_gem_object *obj = intel_fb_obj(old_plane_state->hw.fb);
if (!obj)
return;
if (state->rps_interactive) {
intel_rps_mark_interactive(&dev_priv->gt.rps, false);
state->rps_interactive = false;
}
/* Should only be called after a successful intel_prepare_plane_fb()! */
intel_plane_unpin_fb(old_plane_state);
}
static const struct drm_plane_helper_funcs intel_plane_helper_funcs = {
.prepare_fb = intel_prepare_plane_fb,
.cleanup_fb = intel_cleanup_plane_fb,

View File

@ -68,7 +68,6 @@
#include "gem/i915_gem_lmem.h"
#include "gem/i915_gem_object.h"
#include "gt/intel_rps.h"
#include "gt/gen8_ppgtt.h"
#include "g4x_dp.h"
@ -10479,64 +10478,6 @@ static int intel_atomic_commit(struct drm_device *dev,
return 0;
}
struct wait_rps_boost {
struct wait_queue_entry wait;
struct drm_crtc *crtc;
struct i915_request *request;
};
static int do_rps_boost(struct wait_queue_entry *_wait,
unsigned mode, int sync, void *key)
{
struct wait_rps_boost *wait = container_of(_wait, typeof(*wait), wait);
struct i915_request *rq = wait->request;
/*
* If we missed the vblank, but the request is already running it
* is reasonable to assume that it will complete before the next
* vblank without our intervention, so leave RPS alone.
*/
if (!i915_request_started(rq))
intel_rps_boost(rq);
i915_request_put(rq);
drm_crtc_vblank_put(wait->crtc);
list_del(&wait->wait.entry);
kfree(wait);
return 1;
}
static void add_rps_boost_after_vblank(struct drm_crtc *crtc,
struct dma_fence *fence)
{
struct wait_rps_boost *wait;
if (!dma_fence_is_i915(fence))
return;
if (DISPLAY_VER(to_i915(crtc->dev)) < 6)
return;
if (drm_crtc_vblank_get(crtc))
return;
wait = kmalloc(sizeof(*wait), GFP_KERNEL);
if (!wait) {
drm_crtc_vblank_put(crtc);
return;
}
wait->request = to_request(dma_fence_get(fence));
wait->crtc = crtc;
wait->wait.func = do_rps_boost;
wait->wait.flags = 0;
add_wait_queue(drm_crtc_vblank_waitqueue(crtc), &wait->wait);
}
int intel_plane_pin_fb(struct intel_plane_state *plane_state)
{
struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane);
@ -10603,155 +10544,6 @@ void intel_plane_unpin_fb(struct intel_plane_state *old_plane_state)
}
}
/**
* intel_prepare_plane_fb - Prepare fb for usage on plane
* @_plane: drm plane to prepare for
* @_new_plane_state: the plane state being prepared
*
* Prepares a framebuffer for usage on a display plane. Generally this
* involves pinning the underlying object and updating the frontbuffer tracking
* bits. Some older platforms need special physical address handling for
* cursor planes.
*
* Returns 0 on success, negative error code on failure.
*/
int
intel_prepare_plane_fb(struct drm_plane *_plane,
struct drm_plane_state *_new_plane_state)
{
struct i915_sched_attr attr = { .priority = I915_PRIORITY_DISPLAY };
struct intel_plane *plane = to_intel_plane(_plane);
struct intel_plane_state *new_plane_state =
to_intel_plane_state(_new_plane_state);
struct intel_atomic_state *state =
to_intel_atomic_state(new_plane_state->uapi.state);
struct drm_i915_private *dev_priv = to_i915(plane->base.dev);
const struct intel_plane_state *old_plane_state =
intel_atomic_get_old_plane_state(state, plane);
struct drm_i915_gem_object *obj = intel_fb_obj(new_plane_state->hw.fb);
struct drm_i915_gem_object *old_obj = intel_fb_obj(old_plane_state->hw.fb);
int ret;
if (old_obj) {
const struct intel_crtc_state *crtc_state =
intel_atomic_get_new_crtc_state(state,
to_intel_crtc(old_plane_state->hw.crtc));
/* Big Hammer, we also need to ensure that any pending
* MI_WAIT_FOR_EVENT inside a user batch buffer on the
* current scanout is retired before unpinning the old
* framebuffer. Note that we rely on userspace rendering
* into the buffer attached to the pipe they are waiting
* on. If not, userspace generates a GPU hang with IPEHR
* point to the MI_WAIT_FOR_EVENT.
*
* This should only fail upon a hung GPU, in which case we
* can safely continue.
*/
if (intel_crtc_needs_modeset(crtc_state)) {
ret = i915_sw_fence_await_reservation(&state->commit_ready,
old_obj->base.resv, NULL,
false, 0,
GFP_KERNEL);
if (ret < 0)
return ret;
}
}
if (new_plane_state->uapi.fence) { /* explicit fencing */
i915_gem_fence_wait_priority(new_plane_state->uapi.fence,
&attr);
ret = i915_sw_fence_await_dma_fence(&state->commit_ready,
new_plane_state->uapi.fence,
i915_fence_timeout(dev_priv),
GFP_KERNEL);
if (ret < 0)
return ret;
}
if (!obj)
return 0;
ret = intel_plane_pin_fb(new_plane_state);
if (ret)
return ret;
i915_gem_object_wait_priority(obj, 0, &attr);
if (!new_plane_state->uapi.fence) { /* implicit fencing */
struct dma_fence *fence;
ret = i915_sw_fence_await_reservation(&state->commit_ready,
obj->base.resv, NULL,
false,
i915_fence_timeout(dev_priv),
GFP_KERNEL);
if (ret < 0)
goto unpin_fb;
fence = dma_resv_get_excl_unlocked(obj->base.resv);
if (fence) {
add_rps_boost_after_vblank(new_plane_state->hw.crtc,
fence);
dma_fence_put(fence);
}
} else {
add_rps_boost_after_vblank(new_plane_state->hw.crtc,
new_plane_state->uapi.fence);
}
/*
* We declare pageflips to be interactive and so merit a small bias
* towards upclocking to deliver the frame on time. By only changing
* the RPS thresholds to sample more regularly and aim for higher
* clocks we can hopefully deliver low power workloads (like kodi)
* that are not quite steady state without resorting to forcing
* maximum clocks following a vblank miss (see do_rps_boost()).
*/
if (!state->rps_interactive) {
intel_rps_mark_interactive(&dev_priv->gt.rps, true);
state->rps_interactive = true;
}
return 0;
unpin_fb:
intel_plane_unpin_fb(new_plane_state);
return ret;
}
/**
* intel_cleanup_plane_fb - Cleans up an fb after plane use
* @plane: drm plane to clean up for
* @_old_plane_state: the state from the previous modeset
*
* Cleans up a framebuffer that has just been removed from a plane.
*/
void
intel_cleanup_plane_fb(struct drm_plane *plane,
struct drm_plane_state *_old_plane_state)
{
struct intel_plane_state *old_plane_state =
to_intel_plane_state(_old_plane_state);
struct intel_atomic_state *state =
to_intel_atomic_state(old_plane_state->uapi.state);
struct drm_i915_private *dev_priv = to_i915(plane->dev);
struct drm_i915_gem_object *obj = intel_fb_obj(old_plane_state->hw.fb);
if (!obj)
return;
if (state->rps_interactive) {
intel_rps_mark_interactive(&dev_priv->gt.rps, false);
state->rps_interactive = false;
}
/* Should only be called after a successful intel_prepare_plane_fb()! */
intel_plane_unpin_fb(old_plane_state);
}
/**
* intel_plane_destroy - destroy a plane
* @plane: plane to destroy

View File

@ -585,10 +585,6 @@ void intel_unpin_fb_vma(struct i915_vma *vma, unsigned long flags);
struct drm_framebuffer *
intel_framebuffer_create(struct drm_i915_gem_object *obj,
struct drm_mode_fb_cmd2 *mode_cmd);
int intel_prepare_plane_fb(struct drm_plane *plane,
struct drm_plane_state *new_state);
void intel_cleanup_plane_fb(struct drm_plane *plane,
struct drm_plane_state *old_state);
void assert_pch_transcoder_disabled(struct drm_i915_private *dev_priv,
enum pipe pipe);