microsoft
/
WSL2-Linux-Kernel
зеркало из https://github.com/microsoft/WSL2-Linux-Kernel.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

drm/i915: Split i915_gem_timeline into individual timelines 

We need to move to a more flexible timeline that doesn't assume one
fence context per engine, and so allow for a single timeline to be used
across a combination of engines. This means that preallocating a fence
context per engine is now a hindrance, and so we want to introduce the
singular timeline. From the code perspective, this has the notable
advantage of clearing up a lot of mirky semantics and some clumsy
pointer chasing.

By splitting the timeline up into a single entity rather than an array
of per-engine timelines, we can realise the goal of the previous patch
of tracking the timeline alongside the ring.

v2: Tweak wait_for_idle to stop the compiling thinking that ret may be
uninitialised.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180502163839.3248-2-chris@chris-wilson.co.uk
This commit is contained in:
Chris Wilson 2018-05-02 17:38:39 +01:00
Родитель 65fcb8064d
Коммит a89d1f921c
24 изменённых файлов: 397 добавлений и 582 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

2
drivers/gpu/drm/i915/Makefile
Просмотреть файл

@ -71,11 +71,11 @@ i915-y += i915_cmd_parser.o \
i915_gem_shrinker.o \
i915_gem_stolen.o \
i915_gem_tiling.o \
i915_gem_timeline.o \
i915_gem_userptr.o \
i915_gemfs.o \
i915_query.o \
i915_request.o \
i915_timeline.o \
i915_trace_points.o \
i915_vma.o \
intel_breadcrumbs.o \

4
drivers/gpu/drm/i915/i915_drv.h
Просмотреть файл

@ -72,10 +72,10 @@
#include "i915_gem_fence_reg.h"
#include "i915_gem_object.h"
#include "i915_gem_gtt.h"
#include "i915_gem_timeline.h"
#include "i915_gpu_error.h"
#include "i915_request.h"
#include "i915_scheduler.h"
#include "i915_timeline.h"
#include "i915_vma.h"
#include "intel_gvt.h"
@ -2059,8 +2059,6 @@ struct drm_i915_private {
void (*resume)(struct drm_i915_private *);
void (*cleanup_engine)(struct intel_engine_cs *engine);
struct i915_gem_timeline execution_timeline;
struct i915_gem_timeline legacy_timeline;
struct list_head timelines;
struct list_head active_rings;

91
drivers/gpu/drm/i915/i915_gem.c
Просмотреть файл

@ -162,7 +162,7 @@ static u32 __i915_gem_park(struct drm_i915_private *i915)
synchronize_irq(i915->drm.irq);
intel_engines_park(i915);
i915_gem_timelines_park(i915);
i915_timelines_park(i915);
i915_pmu_gt_parked(i915);
@ -2977,8 +2977,8 @@ i915_gem_find_active_request(struct intel_engine_cs *engine)
* extra delay for a recent interrupt is pointless. Hence, we do
* not need an engine->irq_seqno_barrier() before the seqno reads.
*/
spin_lock_irqsave(&engine->timeline->lock, flags);
list_for_each_entry(request, &engine->timeline->requests, link) {
spin_lock_irqsave(&engine->timeline.lock, flags);
list_for_each_entry(request, &engine->timeline.requests, link) {
if (__i915_request_completed(request, request->global_seqno))
continue;
@ -2989,7 +2989,7 @@ i915_gem_find_active_request(struct intel_engine_cs *engine)
active = request;
break;
}
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
return active;
}
@ -3110,15 +3110,15 @@ static void engine_skip_context(struct i915_request *request)
{
struct intel_engine_cs *engine = request->engine;
struct i915_gem_context *hung_ctx = request->ctx;
struct intel_timeline *timeline = request->timeline;
struct i915_timeline *timeline = request->timeline;
unsigned long flags;
GEM_BUG_ON(timeline == engine->timeline);
GEM_BUG_ON(timeline == &engine->timeline);
spin_lock_irqsave(&engine->timeline->lock, flags);
spin_lock_irqsave(&engine->timeline.lock, flags);
spin_lock(&timeline->lock);
list_for_each_entry_continue(request, &engine->timeline->requests, link)
list_for_each_entry_continue(request, &engine->timeline.requests, link)
if (request->ctx == hung_ctx)
skip_request(request);
@ -3126,7 +3126,7 @@ static void engine_skip_context(struct i915_request *request)
skip_request(request);
spin_unlock(&timeline->lock);
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
}
/* Returns the request if it was guilty of the hang */
@ -3183,11 +3183,11 @@ i915_gem_reset_request(struct intel_engine_cs *engine,
dma_fence_set_error(&request->fence, -EAGAIN);
/* Rewind the engine to replay the incomplete rq */
spin_lock_irq(&engine->timeline->lock);
spin_lock_irq(&engine->timeline.lock);
request = list_prev_entry(request, link);
if (&request->link == &engine->timeline->requests)
if (&request->link == &engine->timeline.requests)
request = NULL;
spin_unlock_irq(&engine->timeline->lock);
spin_unlock_irq(&engine->timeline.lock);
}
}
@ -3300,10 +3300,10 @@ static void nop_complete_submit_request(struct i915_request *request)
request->fence.context, request->fence.seqno);
dma_fence_set_error(&request->fence, -EIO);
spin_lock_irqsave(&request->engine->timeline->lock, flags);
spin_lock_irqsave(&request->engine->timeline.lock, flags);
__i915_request_submit(request);
intel_engine_init_global_seqno(request->engine, request->global_seqno);
spin_unlock_irqrestore(&request->engine->timeline->lock, flags);
spin_unlock_irqrestore(&request->engine->timeline.lock, flags);
}
void i915_gem_set_wedged(struct drm_i915_private *i915)
@ -3372,10 +3372,10 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
* (lockless) lookup doesn't try and wait upon the request as we
* reset it.
*/
spin_lock_irqsave(&engine->timeline->lock, flags);
spin_lock_irqsave(&engine->timeline.lock, flags);
intel_engine_init_global_seqno(engine,
intel_engine_last_submit(engine));
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
i915_gem_reset_finish_engine(engine);
}
@ -3387,8 +3387,7 @@ void i915_gem_set_wedged(struct drm_i915_private *i915)
bool i915_gem_unset_wedged(struct drm_i915_private *i915)
{
struct i915_gem_timeline *tl;
int i;
struct i915_timeline *tl;
lockdep_assert_held(&i915->drm.struct_mutex);
if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
@ -3407,10 +3406,9 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
* No more can be submitted until we reset the wedged bit.
*/
list_for_each_entry(tl, &i915->gt.timelines, link) {
for (i = 0; i < ARRAY_SIZE(tl->engine); i++) {
struct i915_request *rq;
rq = i915_gem_active_peek(&tl->engine[i].last_request,
rq = i915_gem_active_peek(&tl->last_request,
&i915->drm.struct_mutex);
if (!rq)
continue;
@ -3430,7 +3428,6 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
MAX_SCHEDULE_TIMEOUT) < 0)
return false;
}
}
i915_retire_requests(i915);
GEM_BUG_ON(i915->gt.active_requests);
@ -3734,17 +3731,9 @@ i915_gem_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
return ret;
}
static int wait_for_timeline(struct i915_gem_timeline *tl, unsigned int flags)
static int wait_for_timeline(struct i915_timeline *tl, unsigned int flags)
{
int ret, i;
for (i = 0; i < ARRAY_SIZE(tl->engine); i++) {
ret = i915_gem_active_wait(&tl->engine[i].last_request, flags);
if (ret)
return ret;
}
return 0;
return i915_gem_active_wait(&tl->last_request, flags);
}
static int wait_for_engines(struct drm_i915_private *i915)
@ -3762,30 +3751,37 @@ static int wait_for_engines(struct drm_i915_private *i915)
int i915_gem_wait_for_idle(struct drm_i915_private *i915, unsigned int flags)
{
int ret;
/* If the device is asleep, we have no requests outstanding */
if (!READ_ONCE(i915->gt.awake))
return 0;
if (flags & I915_WAIT_LOCKED) {
struct i915_gem_timeline *tl;
struct i915_timeline *tl;
int err;
lockdep_assert_held(&i915->drm.struct_mutex);
list_for_each_entry(tl, &i915->gt.timelines, link) {
ret = wait_for_timeline(tl, flags);
if (ret)
return ret;
err = wait_for_timeline(tl, flags);
if (err)
return err;
}
i915_retire_requests(i915);
ret = wait_for_engines(i915);
return wait_for_engines(i915);
} else {
ret = wait_for_timeline(&i915->gt.execution_timeline, flags);
struct intel_engine_cs *engine;
enum intel_engine_id id;
int err;
for_each_engine(engine, i915, id) {
err = wait_for_timeline(&engine->timeline, flags);
if (err)
return err;
}
return ret;
return 0;
}
}
static void __i915_gem_object_flush_for_display(struct drm_i915_gem_object *obj)
@ -4954,7 +4950,7 @@ static void assert_kernel_context_is_current(struct drm_i915_private *i915)
enum intel_engine_id id;
for_each_engine(engine, i915, id) {
GEM_BUG_ON(__i915_gem_active_peek(&engine->timeline->last_request));
GEM_BUG_ON(__i915_gem_active_peek(&engine->timeline.last_request));
GEM_BUG_ON(engine->last_retired_context != kernel_context);
}
}
@ -5603,12 +5599,6 @@ int i915_gem_init_early(struct drm_i915_private *dev_priv)
INIT_LIST_HEAD(&dev_priv->gt.timelines);
INIT_LIST_HEAD(&dev_priv->gt.active_rings);
mutex_lock(&dev_priv->drm.struct_mutex);
err = i915_gem_timeline_init__global(dev_priv);
mutex_unlock(&dev_priv->drm.struct_mutex);
if (err)
goto err_priorities;
i915_gem_init__mm(dev_priv);
INIT_DELAYED_WORK(&dev_priv->gt.retire_work,
@ -5628,8 +5618,6 @@ int i915_gem_init_early(struct drm_i915_private *dev_priv)
return 0;
err_priorities:
kmem_cache_destroy(dev_priv->priorities);
err_dependencies:
kmem_cache_destroy(dev_priv->dependencies);
err_requests:
@ -5650,12 +5638,7 @@ void i915_gem_cleanup_early(struct drm_i915_private *dev_priv)
GEM_BUG_ON(!llist_empty(&dev_priv->mm.free_list));
GEM_BUG_ON(atomic_read(&dev_priv->mm.free_count));
WARN_ON(dev_priv->mm.object_count);
mutex_lock(&dev_priv->drm.struct_mutex);
i915_gem_timeline_fini(&dev_priv->gt.legacy_timeline);
i915_gem_timeline_fini(&dev_priv->gt.execution_timeline);
WARN_ON(!list_empty(&dev_priv->gt.timelines));
mutex_unlock(&dev_priv->drm.struct_mutex);
kmem_cache_destroy(dev_priv->priorities);
kmem_cache_destroy(dev_priv->dependencies);

48
drivers/gpu/drm/i915/i915_gem_context.c
Просмотреть файл

@ -122,7 +122,6 @@ static void i915_gem_context_free(struct i915_gem_context *ctx)
lockdep_assert_held(&ctx->i915->drm.struct_mutex);
GEM_BUG_ON(!i915_gem_context_is_closed(ctx));
i915_gem_timeline_free(ctx->timeline);
i915_ppgtt_put(ctx->ppgtt);
for (n = 0; n < ARRAY_SIZE(ctx->__engine); n++) {
@ -377,18 +376,6 @@ i915_gem_create_context(struct drm_i915_private *dev_priv,
ctx->desc_template = default_desc_template(dev_priv, ppgtt);
}
if (HAS_EXECLISTS(dev_priv)) {
struct i915_gem_timeline *timeline;
timeline = i915_gem_timeline_create(dev_priv, ctx->name);
if (IS_ERR(timeline)) {
__destroy_hw_context(ctx, file_priv);
return ERR_CAST(timeline);
}
ctx->timeline = timeline;
}
trace_i915_context_create(ctx);
return ctx;
@ -590,19 +577,29 @@ void i915_gem_context_close(struct drm_file *file)
idr_destroy(&file_priv->context_idr);
}
static struct i915_request *
last_request_on_engine(struct i915_timeline *timeline,
struct intel_engine_cs *engine)
{
struct i915_request *rq;
if (timeline == &engine->timeline)
return NULL;
rq = i915_gem_active_raw(&timeline->last_request,
&engine->i915->drm.struct_mutex);
if (rq && rq->engine == engine)
return rq;
return NULL;
}
static bool engine_has_idle_kernel_context(struct intel_engine_cs *engine)
{
struct i915_gem_timeline *timeline;
struct i915_timeline *timeline;
list_for_each_entry(timeline, &engine->i915->gt.timelines, link) {
struct intel_timeline *tl;
if (timeline == &engine->i915->gt.execution_timeline)
continue;
tl = &timeline->engine[engine->id];
if (i915_gem_active_peek(&tl->last_request,
&engine->i915->drm.struct_mutex))
if (last_request_on_engine(timeline, engine))
return false;
}
@ -612,7 +609,7 @@ static bool engine_has_idle_kernel_context(struct intel_engine_cs *engine)
int i915_gem_switch_to_kernel_context(struct drm_i915_private *dev_priv)
{
struct intel_engine_cs *engine;
struct i915_gem_timeline *timeline;
struct i915_timeline *timeline;
enum intel_engine_id id;
lockdep_assert_held(&dev_priv->drm.struct_mutex);
@ -632,11 +629,8 @@ int i915_gem_switch_to_kernel_context(struct drm_i915_private *dev_priv)
/* Queue this switch after all other activity */
list_for_each_entry(timeline, &dev_priv->gt.timelines, link) {
struct i915_request *prev;
struct intel_timeline *tl;
tl = &timeline->engine[engine->id];
prev = i915_gem_active_raw(&tl->last_request,
&dev_priv->drm.struct_mutex);
prev = last_request_on_engine(timeline, engine);
if (prev)
i915_sw_fence_await_sw_fence_gfp(&rq->submit,
&prev->submit,

2
drivers/gpu/drm/i915/i915_gem_context.h
Просмотреть файл

@ -58,8 +58,6 @@ struct i915_gem_context {
/** file_priv: owning file descriptor */
struct drm_i915_file_private *file_priv;
struct i915_gem_timeline *timeline;
/**
* @ppgtt: unique address space (GTT)
*

3
drivers/gpu/drm/i915/i915_gem_gtt.h
Просмотреть файл

@ -38,10 +38,9 @@
#include <linux/mm.h>
#include <linux/pagevec.h>
#include "i915_gem_timeline.h"
#include "i915_request.h"
#include "i915_selftest.h"
#include "i915_timeline.h"
#define I915_GTT_PAGE_SIZE_4K BIT(12)
#define I915_GTT_PAGE_SIZE_64K BIT(16)

198
drivers/gpu/drm/i915/i915_gem_timeline.c
Просмотреть файл

@ -1,198 +0,0 @@
/*
* Copyright © 2016 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
*
*/
#include "i915_drv.h"
#include "i915_syncmap.h"
static void __intel_timeline_init(struct intel_timeline *tl,
struct i915_gem_timeline *parent,
u64 context,
struct lock_class_key *lockclass,
const char *lockname)
{
tl->fence_context = context;
tl->common = parent;
spin_lock_init(&tl->lock);
lockdep_set_class_and_name(&tl->lock, lockclass, lockname);
init_request_active(&tl->last_request, NULL);
INIT_LIST_HEAD(&tl->requests);
i915_syncmap_init(&tl->sync);
}
static void __intel_timeline_fini(struct intel_timeline *tl)
{
GEM_BUG_ON(!list_empty(&tl->requests));
i915_syncmap_free(&tl->sync);
}
static int __i915_gem_timeline_init(struct drm_i915_private *i915,
struct i915_gem_timeline *timeline,
const char *name,
struct lock_class_key *lockclass,
const char *lockname)
{
unsigned int i;
u64 fences;
lockdep_assert_held(&i915->drm.struct_mutex);
/*
* Ideally we want a set of engines on a single leaf as we expect
* to mostly be tracking synchronisation between engines. It is not
* a huge issue if this is not the case, but we may want to mitigate
* any page crossing penalties if they become an issue.
*/
BUILD_BUG_ON(KSYNCMAP < I915_NUM_ENGINES);
timeline->i915 = i915;
timeline->name = kstrdup(name ?: "[kernel]", GFP_KERNEL);
if (!timeline->name)
return -ENOMEM;
list_add(&timeline->link, &i915->gt.timelines);
/* Called during early_init before we know how many engines there are */
fences = dma_fence_context_alloc(ARRAY_SIZE(timeline->engine));
for (i = 0; i < ARRAY_SIZE(timeline->engine); i++)
__intel_timeline_init(&timeline->engine[i],
timeline, fences++,
lockclass, lockname);
return 0;
}
int i915_gem_timeline_init(struct drm_i915_private *i915,
struct i915_gem_timeline *timeline,
const char *name)
{
static struct lock_class_key class;
return __i915_gem_timeline_init(i915, timeline, name,
&class, "&timeline->lock");
}
int i915_gem_timeline_init__global(struct drm_i915_private *i915)
{
static struct lock_class_key class1, class2;
int err;
err = __i915_gem_timeline_init(i915,
&i915->gt.execution_timeline,
"[execution]", &class1,
"i915_execution_timeline");
if (err)
return err;
err = __i915_gem_timeline_init(i915,
&i915->gt.legacy_timeline,
"[global]", &class2,
"i915_global_timeline");
if (err)
goto err_exec_timeline;
return 0;
err_exec_timeline:
i915_gem_timeline_fini(&i915->gt.execution_timeline);
return err;
}
/**
* i915_gem_timelines_park - called when the driver idles
* @i915: the drm_i915_private device
*
* When the driver is completely idle, we know that all of our sync points
* have been signaled and our tracking is then entirely redundant. Any request
* to wait upon an older sync point will be completed instantly as we know
* the fence is signaled and therefore we will not even look them up in the
* sync point map.
*/
void i915_gem_timelines_park(struct drm_i915_private *i915)
{
struct i915_gem_timeline *timeline;
int i;
lockdep_assert_held(&i915->drm.struct_mutex);
list_for_each_entry(timeline, &i915->gt.timelines, link) {
for (i = 0; i < ARRAY_SIZE(timeline->engine); i++) {
struct intel_timeline *tl = &timeline->engine[i];
/*
* All known fences are completed so we can scrap
* the current sync point tracking and start afresh,
* any attempt to wait upon a previous sync point
* will be skipped as the fence was signaled.
*/
i915_syncmap_free(&tl->sync);
}
}
}
void i915_gem_timeline_fini(struct i915_gem_timeline *timeline)
{
int i;
lockdep_assert_held(&timeline->i915->drm.struct_mutex);
for (i = 0; i < ARRAY_SIZE(timeline->engine); i++)
__intel_timeline_fini(&timeline->engine[i]);
list_del(&timeline->link);
kfree(timeline->name);
}
struct i915_gem_timeline *
i915_gem_timeline_create(struct drm_i915_private *i915, const char *name)
{
struct i915_gem_timeline *timeline;
int err;
timeline = kzalloc(sizeof(*timeline), GFP_KERNEL);
if (!timeline)
return ERR_PTR(-ENOMEM);
err = i915_gem_timeline_init(i915, timeline, name);
if (err) {
kfree(timeline);
return ERR_PTR(err);
}
return timeline;
}
void i915_gem_timeline_free(struct i915_gem_timeline *timeline)
{
if (!timeline)
return;
i915_gem_timeline_fini(timeline);
kfree(timeline);
}
#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
#include "selftests/mock_timeline.c"
#include "selftests/i915_gem_timeline.c"
#endif

4
drivers/gpu/drm/i915/i915_gpu_error.c
Просмотреть файл

@ -1310,7 +1310,7 @@ static void engine_record_requests(struct intel_engine_cs *engine,
count = 0;
request = first;
list_for_each_entry_from(request, &engine->timeline->requests, link)
list_for_each_entry_from(request, &engine->timeline.requests, link)
count++;
if (!count)
return;
@ -1323,7 +1323,7 @@ static void engine_record_requests(struct intel_engine_cs *engine,
count = 0;
request = first;
list_for_each_entry_from(request, &engine->timeline->requests, link) {
list_for_each_entry_from(request, &engine->timeline.requests, link) {
if (count >= ee->num_requests) {
/*
* If the ring request list was changed in

10
drivers/gpu/drm/i915/i915_perf.c
Просмотреть файл

@ -1695,7 +1695,7 @@ static int gen8_switch_to_updated_kernel_context(struct drm_i915_private *dev_pr
const struct i915_oa_config *oa_config)
{
struct intel_engine_cs *engine = dev_priv->engine[RCS];
struct i915_gem_timeline *timeline;
struct i915_timeline *timeline;
struct i915_request *rq;
int ret;
@ -1716,15 +1716,11 @@ static int gen8_switch_to_updated_kernel_context(struct drm_i915_private *dev_pr
/* Queue this switch after all other activity */
list_for_each_entry(timeline, &dev_priv->gt.timelines, link) {
struct i915_request *prev;
struct intel_timeline *tl;
tl = &timeline->engine[engine->id];
prev = i915_gem_active_raw(&tl->last_request,
prev = i915_gem_active_raw(&timeline->last_request,
&dev_priv->drm.struct_mutex);
if (prev)
i915_sw_fence_await_sw_fence_gfp(&rq->submit,
&prev->submit,
GFP_KERNEL);
i915_request_await_dma_fence(rq, &prev->fence);
}
i915_request_add(rq);

68
drivers/gpu/drm/i915/i915_request.c
Просмотреть файл

@ -49,7 +49,7 @@ static const char *i915_fence_get_timeline_name(struct dma_fence *fence)
if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags))
return "signaled";
return to_request(fence)->timeline->common->name;
return to_request(fence)->timeline->name;
}
static bool i915_fence_signaled(struct dma_fence *fence)
@ -199,6 +199,7 @@ i915_sched_node_init(struct i915_sched_node *node)
static int reset_all_global_seqno(struct drm_i915_private *i915, u32 seqno)
{
struct intel_engine_cs *engine;
struct i915_timeline *timeline;
enum intel_engine_id id;
int ret;
@ -213,16 +214,13 @@ static int reset_all_global_seqno(struct drm_i915_private *i915, u32 seqno)
/* If the seqno wraps around, we need to clear the breadcrumb rbtree */
for_each_engine(engine, i915, id) {
struct i915_gem_timeline *timeline;
struct intel_timeline *tl = engine->timeline;
GEM_TRACE("%s seqno %d (current %d) -> %d\n",
engine->name,
tl->seqno,
engine->timeline.seqno,
intel_engine_get_seqno(engine),
seqno);
if (!i915_seqno_passed(seqno, tl->seqno)) {
if (!i915_seqno_passed(seqno, engine->timeline.seqno)) {
/* Flush any waiters before we reuse the seqno */
intel_engine_disarm_breadcrumbs(engine);
GEM_BUG_ON(!list_empty(&engine->breadcrumbs.signals));
@ -230,18 +228,18 @@ static int reset_all_global_seqno(struct drm_i915_private *i915, u32 seqno)
/* Check we are idle before we fiddle with hw state! */
GEM_BUG_ON(!intel_engine_is_idle(engine));
GEM_BUG_ON(i915_gem_active_isset(&engine->timeline->last_request));
GEM_BUG_ON(i915_gem_active_isset(&engine->timeline.last_request));
/* Finally reset hw state */
intel_engine_init_global_seqno(engine, seqno);
tl->seqno = seqno;
list_for_each_entry(timeline, &i915->gt.timelines, link)
memset(timeline->engine[id].global_sync, 0,
sizeof(timeline->engine[id].global_sync));
engine->timeline.seqno = seqno;
}
list_for_each_entry(timeline, &i915->gt.timelines, link)
memset(timeline->global_sync, 0, sizeof(timeline->global_sync));
i915->gt.request_serial = seqno;
return 0;
}
@ -357,10 +355,10 @@ static void __retire_engine_request(struct intel_engine_cs *engine,
local_irq_disable();
spin_lock(&engine->timeline->lock);
GEM_BUG_ON(!list_is_first(&rq->link, &engine->timeline->requests));
spin_lock(&engine->timeline.lock);
GEM_BUG_ON(!list_is_first(&rq->link, &engine->timeline.requests));
list_del_init(&rq->link);
spin_unlock(&engine->timeline->lock);
spin_unlock(&engine->timeline.lock);
spin_lock(&rq->lock);
if (!test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &rq->fence.flags))
@ -397,7 +395,7 @@ static void __retire_engine_upto(struct intel_engine_cs *engine,
return;
do {
tmp = list_first_entry(&engine->timeline->requests,
tmp = list_first_entry(&engine->timeline.requests,
typeof(*tmp), link);
GEM_BUG_ON(tmp->engine != engine);
@ -492,16 +490,16 @@ void i915_request_retire_upto(struct i915_request *rq)
} while (tmp != rq);
}
static u32 timeline_get_seqno(struct intel_timeline *tl)
static u32 timeline_get_seqno(struct i915_timeline *tl)
{
return ++tl->seqno;
}
static void move_to_timeline(struct i915_request *request,
struct intel_timeline *timeline)
struct i915_timeline *timeline)
{
GEM_BUG_ON(request->timeline == request->engine->timeline);
lockdep_assert_held(&request->engine->timeline->lock);
GEM_BUG_ON(request->timeline == &request->engine->timeline);
lockdep_assert_held(&request->engine->timeline.lock);
spin_lock(&request->timeline->lock);
list_move_tail(&request->link, &timeline->requests);
@ -516,15 +514,15 @@ void __i915_request_submit(struct i915_request *request)
GEM_TRACE("%s fence %llx:%d -> global=%d, current %d\n",
engine->name,
request->fence.context, request->fence.seqno,
engine->timeline->seqno + 1,
engine->timeline.seqno + 1,
intel_engine_get_seqno(engine));
GEM_BUG_ON(!irqs_disabled());
lockdep_assert_held(&engine->timeline->lock);
lockdep_assert_held(&engine->timeline.lock);
GEM_BUG_ON(request->global_seqno);
seqno = timeline_get_seqno(engine->timeline);
seqno = timeline_get_seqno(&engine->timeline);
GEM_BUG_ON(!seqno);
GEM_BUG_ON(i915_seqno_passed(intel_engine_get_seqno(engine), seqno));
@ -539,7 +537,7 @@ void __i915_request_submit(struct i915_request *request)
request->ring->vaddr + request->postfix);
/* Transfer from per-context onto the global per-engine timeline */
move_to_timeline(request, engine->timeline);
move_to_timeline(request, &engine->timeline);
trace_i915_request_execute(request);
@ -552,11 +550,11 @@ void i915_request_submit(struct i915_request *request)
unsigned long flags;
/* Will be called from irq-context when using foreign fences. */
spin_lock_irqsave(&engine->timeline->lock, flags);
spin_lock_irqsave(&engine->timeline.lock, flags);
__i915_request_submit(request);
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
}
void __i915_request_unsubmit(struct i915_request *request)
@ -570,17 +568,17 @@ void __i915_request_unsubmit(struct i915_request *request)
intel_engine_get_seqno(engine));
GEM_BUG_ON(!irqs_disabled());
lockdep_assert_held(&engine->timeline->lock);
lockdep_assert_held(&engine->timeline.lock);
/*
* Only unwind in reverse order, required so that the per-context list
* is kept in seqno/ring order.
*/
GEM_BUG_ON(!request->global_seqno);
GEM_BUG_ON(request->global_seqno != engine->timeline->seqno);
GEM_BUG_ON(request->global_seqno != engine->timeline.seqno);
GEM_BUG_ON(i915_seqno_passed(intel_engine_get_seqno(engine),
request->global_seqno));
engine->timeline->seqno--;
engine->timeline.seqno--;
/* We may be recursing from the signal callback of another i915 fence */
spin_lock_nested(&request->lock, SINGLE_DEPTH_NESTING);
@ -607,11 +605,11 @@ void i915_request_unsubmit(struct i915_request *request)
unsigned long flags;
/* Will be called from irq-context when using foreign fences. */
spin_lock_irqsave(&engine->timeline->lock, flags);
spin_lock_irqsave(&engine->timeline.lock, flags);
__i915_request_unsubmit(request);
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
}
static int __i915_sw_fence_call
@ -764,7 +762,7 @@ i915_request_alloc(struct intel_engine_cs *engine, struct i915_gem_context *ctx)
rq->ctx = ctx;
rq->ring = ring;
rq->timeline = ring->timeline;
GEM_BUG_ON(rq->timeline == engine->timeline);
GEM_BUG_ON(rq->timeline == &engine->timeline);
spin_lock_init(&rq->lock);
dma_fence_init(&rq->fence,
@ -929,7 +927,7 @@ i915_request_await_dma_fence(struct i915_request *rq, struct dma_fence *fence)
/* Squash repeated waits to the same timelines */
if (fence->context != rq->i915->mm.unordered_timeline &&
intel_timeline_sync_is_later(rq->timeline, fence))
i915_timeline_sync_is_later(rq->timeline, fence))
continue;
if (dma_fence_is_i915(fence))
@ -943,7 +941,7 @@ i915_request_await_dma_fence(struct i915_request *rq, struct dma_fence *fence)
/* Record the latest fence used against each timeline */
if (fence->context != rq->i915->mm.unordered_timeline)
intel_timeline_sync_set(rq->timeline, fence);
i915_timeline_sync_set(rq->timeline, fence);
} while (--nchild);
return 0;
@ -1020,7 +1018,7 @@ void __i915_request_add(struct i915_request *request, bool flush_caches)
{
struct intel_engine_cs *engine = request->engine;
struct intel_ring *ring = request->ring;
struct intel_timeline *timeline = request->timeline;
struct i915_timeline *timeline = request->timeline;
struct i915_request *prev;
u32 *cs;
int err;

3
drivers/gpu/drm/i915/i915_request.h
Просмотреть файл

@ -37,6 +37,7 @@
struct drm_file;
struct drm_i915_gem_object;
struct i915_request;
struct i915_timeline;
struct intel_wait {
struct rb_node node;
@ -95,7 +96,7 @@ struct i915_request {
struct i915_gem_context *ctx;
struct intel_engine_cs *engine;
struct intel_ring *ring;
struct intel_timeline *timeline;
struct i915_timeline *timeline;
struct intel_signal_node signaling;
/*

105
drivers/gpu/drm/i915/i915_timeline.c Normal file
Просмотреть файл

@ -0,0 +1,105 @@
/*
* SPDX-License-Identifier: MIT
*
* Copyright © 2016-2018 Intel Corporation
*/
#include "i915_drv.h"
#include "i915_timeline.h"
#include "i915_syncmap.h"
void i915_timeline_init(struct drm_i915_private *i915,
struct i915_timeline *timeline,
const char *name)
{
lockdep_assert_held(&i915->drm.struct_mutex);
/*
* Ideally we want a set of engines on a single leaf as we expect
* to mostly be tracking synchronisation between engines. It is not
* a huge issue if this is not the case, but we may want to mitigate
* any page crossing penalties if they become an issue.
*/
BUILD_BUG_ON(KSYNCMAP < I915_NUM_ENGINES);
timeline->name = name;
list_add(&timeline->link, &i915->gt.timelines);
/* Called during early_init before we know how many engines there are */
timeline->fence_context = dma_fence_context_alloc(1);
spin_lock_init(&timeline->lock);
init_request_active(&timeline->last_request, NULL);
INIT_LIST_HEAD(&timeline->requests);
i915_syncmap_init(&timeline->sync);
}
/**
* i915_timelines_park - called when the driver idles
* @i915: the drm_i915_private device
*
* When the driver is completely idle, we know that all of our sync points
* have been signaled and our tracking is then entirely redundant. Any request
* to wait upon an older sync point will be completed instantly as we know
* the fence is signaled and therefore we will not even look them up in the
* sync point map.
*/
void i915_timelines_park(struct drm_i915_private *i915)
{
struct i915_timeline *timeline;
lockdep_assert_held(&i915->drm.struct_mutex);
list_for_each_entry(timeline, &i915->gt.timelines, link) {
/*
* All known fences are completed so we can scrap
* the current sync point tracking and start afresh,
* any attempt to wait upon a previous sync point
* will be skipped as the fence was signaled.
*/
i915_syncmap_free(&timeline->sync);
}
}
void i915_timeline_fini(struct i915_timeline *timeline)
{
GEM_BUG_ON(!list_empty(&timeline->requests));
i915_syncmap_free(&timeline->sync);
list_del(&timeline->link);
}
struct i915_timeline *
i915_timeline_create(struct drm_i915_private *i915, const char *name)
{
struct i915_timeline *timeline;
timeline = kzalloc(sizeof(*timeline), GFP_KERNEL);
if (!timeline)
return ERR_PTR(-ENOMEM);
i915_timeline_init(i915, timeline, name);
kref_init(&timeline->kref);
return timeline;
}
void __i915_timeline_free(struct kref *kref)
{
struct i915_timeline *timeline =
container_of(kref, typeof(*timeline), kref);
i915_timeline_fini(timeline);
kfree(timeline);
}
#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
#include "selftests/mock_timeline.c"
#include "selftests/i915_timeline.c"
#endif

57
drivers/gpu/drm/i915/i915_gem_timeline.h → drivers/gpu/drm/i915/i915_timeline.h
Просмотреть файл

@ -22,18 +22,17 @@
*
*/
#ifndef I915_GEM_TIMELINE_H
#define I915_GEM_TIMELINE_H
#ifndef I915_TIMELINE_H
#define I915_TIMELINE_H
#include <linux/list.h>
#include <linux/kref.h>
#include "i915_request.h"
#include "i915_syncmap.h"
#include "i915_utils.h"
struct i915_gem_timeline;
struct intel_timeline {
struct i915_timeline {
u64 fence_context;
u32 seqno;
@ -71,51 +70,57 @@ struct intel_timeline {
*/
u32 global_sync[I915_NUM_ENGINES];
struct i915_gem_timeline *common;
};
struct i915_gem_timeline {
struct list_head link;
struct drm_i915_private *i915;
const char *name;
struct intel_timeline engine[I915_NUM_ENGINES];
struct kref kref;
};
int i915_gem_timeline_init(struct drm_i915_private *i915,
struct i915_gem_timeline *tl,
void i915_timeline_init(struct drm_i915_private *i915,
struct i915_timeline *tl,
const char *name);
int i915_gem_timeline_init__global(struct drm_i915_private *i915);
void i915_gem_timelines_park(struct drm_i915_private *i915);
void i915_gem_timeline_fini(struct i915_gem_timeline *tl);
void i915_timeline_fini(struct i915_timeline *tl);
struct i915_gem_timeline *
i915_gem_timeline_create(struct drm_i915_private *i915, const char *name);
void i915_gem_timeline_free(struct i915_gem_timeline *timeline);
struct i915_timeline *
i915_timeline_create(struct drm_i915_private *i915, const char *name);
static inline int __intel_timeline_sync_set(struct intel_timeline *tl,
static inline struct i915_timeline *
i915_timeline_get(struct i915_timeline *timeline)
{
kref_get(&timeline->kref);
return timeline;
}
void __i915_timeline_free(struct kref *kref);
static inline void i915_timeline_put(struct i915_timeline *timeline)
{
kref_put(&timeline->kref, __i915_timeline_free);
}
static inline int __i915_timeline_sync_set(struct i915_timeline *tl,
u64 context, u32 seqno)
{
return i915_syncmap_set(&tl->sync, context, seqno);
}
static inline int intel_timeline_sync_set(struct intel_timeline *tl,
static inline int i915_timeline_sync_set(struct i915_timeline *tl,
const struct dma_fence *fence)
{
return __intel_timeline_sync_set(tl, fence->context, fence->seqno);
return __i915_timeline_sync_set(tl, fence->context, fence->seqno);
}
static inline bool __intel_timeline_sync_is_later(struct intel_timeline *tl,
static inline bool __i915_timeline_sync_is_later(struct i915_timeline *tl,
u64 context, u32 seqno)
{
return i915_syncmap_is_later(&tl->sync, context, seqno);
}
static inline bool intel_timeline_sync_is_later(struct intel_timeline *tl,
static inline bool i915_timeline_sync_is_later(struct i915_timeline *tl,
const struct dma_fence *fence)
{
return __intel_timeline_sync_is_later(tl, fence->context, fence->seqno);
return __i915_timeline_sync_is_later(tl, fence->context, fence->seqno);
}
void i915_timelines_park(struct drm_i915_private *i915);
#endif

27
drivers/gpu/drm/i915/intel_engine_cs.c
Просмотреть файл

@ -451,12 +451,6 @@ void intel_engine_init_global_seqno(struct intel_engine_cs *engine, u32 seqno)
GEM_BUG_ON(intel_engine_get_seqno(engine) != seqno);
}
static void intel_engine_init_timeline(struct intel_engine_cs *engine)
{
engine->timeline =
&engine->i915->gt.execution_timeline.engine[engine->id];
}
static void intel_engine_init_batch_pool(struct intel_engine_cs *engine)
{
i915_gem_batch_pool_init(&engine->batch_pool, engine);
@ -508,8 +502,9 @@ static void intel_engine_init_execlist(struct intel_engine_cs *engine)
*/
void intel_engine_setup_common(struct intel_engine_cs *engine)
{
i915_timeline_init(engine->i915, &engine->timeline, engine->name);
intel_engine_init_execlist(engine);
intel_engine_init_timeline(engine);
intel_engine_init_hangcheck(engine);
intel_engine_init_batch_pool(engine);
intel_engine_init_cmd_parser(engine);
@ -751,6 +746,8 @@ void intel_engine_cleanup_common(struct intel_engine_cs *engine)
if (engine->i915->preempt_context)
intel_context_unpin(engine->i915->preempt_context, engine);
intel_context_unpin(engine->i915->kernel_context, engine);
i915_timeline_fini(&engine->timeline);
}
u64 intel_engine_get_active_head(const struct intel_engine_cs *engine)
@ -1003,7 +1000,7 @@ bool intel_engine_has_kernel_context(const struct intel_engine_cs *engine)
* the last request that remains in the timeline. When idle, it is
* the last executed context as tracked by retirement.
*/
rq = __i915_gem_active_peek(&engine->timeline->last_request);
rq = __i915_gem_active_peek(&engine->timeline.last_request);
if (rq)
return rq->ctx == kernel_context;
else
@ -1335,14 +1332,14 @@ void intel_engine_dump(struct intel_engine_cs *engine,
drm_printf(m, "\tRequests:\n");
rq = list_first_entry(&engine->timeline->requests,
rq = list_first_entry(&engine->timeline.requests,
struct i915_request, link);
if (&rq->link != &engine->timeline->requests)
if (&rq->link != &engine->timeline.requests)
print_request(m, rq, "\t\tfirst ");
rq = list_last_entry(&engine->timeline->requests,
rq = list_last_entry(&engine->timeline.requests,
struct i915_request, link);
if (&rq->link != &engine->timeline->requests)
if (&rq->link != &engine->timeline.requests)
print_request(m, rq, "\t\tlast ");
rq = i915_gem_find_active_request(engine);
@ -1374,11 +1371,11 @@ void intel_engine_dump(struct intel_engine_cs *engine,
drm_printf(m, "\tDevice is asleep; skipping register dump\n");
}
spin_lock_irq(&engine->timeline->lock);
spin_lock_irq(&engine->timeline.lock);
last = NULL;
count = 0;
list_for_each_entry(rq, &engine->timeline->requests, link) {
list_for_each_entry(rq, &engine->timeline.requests, link) {
if (count++ < MAX_REQUESTS_TO_SHOW - 1)
print_request(m, rq, "\t\tE ");
else
@ -1416,7 +1413,7 @@ void intel_engine_dump(struct intel_engine_cs *engine,
print_request(m, last, "\t\tQ ");
}
spin_unlock_irq(&engine->timeline->lock);
spin_unlock_irq(&engine->timeline.lock);
spin_lock_irq(&b->rb_lock);
for (rb = rb_first(&b->waiters); rb; rb = rb_next(rb)) {

4
drivers/gpu/drm/i915/intel_guc_submission.c
Просмотреть файл

@ -679,7 +679,7 @@ static void guc_dequeue(struct intel_engine_cs *engine)
bool submit = false;
struct rb_node *rb;
spin_lock_irq(&engine->timeline->lock);
spin_lock_irq(&engine->timeline.lock);
rb = execlists->first;
GEM_BUG_ON(rb_first(&execlists->queue) != rb);
@ -750,7 +750,7 @@ done:
GEM_BUG_ON(execlists->first && !port_isset(execlists->port));
unlock:
spin_unlock_irq(&engine->timeline->lock);
spin_unlock_irq(&engine->timeline.lock);
}
static void guc_submission_tasklet(unsigned long data)

48
drivers/gpu/drm/i915/intel_lrc.c
Просмотреть файл

@ -331,10 +331,10 @@ static void __unwind_incomplete_requests(struct intel_engine_cs *engine)
struct i915_priolist *uninitialized_var(p);
int last_prio = I915_PRIORITY_INVALID;
lockdep_assert_held(&engine->timeline->lock);
lockdep_assert_held(&engine->timeline.lock);
list_for_each_entry_safe_reverse(rq, rn,
&engine->timeline->requests,
&engine->timeline.requests,
link) {
if (i915_request_completed(rq))
return;
@ -358,9 +358,9 @@ execlists_unwind_incomplete_requests(struct intel_engine_execlists *execlists)
struct intel_engine_cs *engine =
container_of(execlists, typeof(*engine), execlists);
spin_lock_irq(&engine->timeline->lock);
spin_lock_irq(&engine->timeline.lock);
__unwind_incomplete_requests(engine);
spin_unlock_irq(&engine->timeline->lock);
spin_unlock_irq(&engine->timeline.lock);
}
static inline void
@ -584,7 +584,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
* and context switches) submission.
*/
spin_lock_irq(&engine->timeline->lock);
spin_lock_irq(&engine->timeline.lock);
rb = execlists->first;
GEM_BUG_ON(rb_first(&execlists->queue) != rb);
@ -744,7 +744,7 @@ done:
GEM_BUG_ON(execlists->first && !port_isset(execlists->port));
unlock:
spin_unlock_irq(&engine->timeline->lock);
spin_unlock_irq(&engine->timeline.lock);
if (submit) {
execlists_user_begin(execlists, execlists->port);
@ -894,10 +894,10 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
execlists_cancel_port_requests(execlists);
reset_irq(engine);
spin_lock(&engine->timeline->lock);
spin_lock(&engine->timeline.lock);
/* Mark all executing requests as skipped. */
list_for_each_entry(rq, &engine->timeline->requests, link) {
list_for_each_entry(rq, &engine->timeline.requests, link) {
GEM_BUG_ON(!rq->global_seqno);
if (!i915_request_completed(rq))
dma_fence_set_error(&rq->fence, -EIO);
@ -929,7 +929,7 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
execlists->first = NULL;
GEM_BUG_ON(port_isset(execlists->port));
spin_unlock(&engine->timeline->lock);
spin_unlock(&engine->timeline.lock);
local_irq_restore(flags);
}
@ -1167,7 +1167,7 @@ static void execlists_submit_request(struct i915_request *request)
unsigned long flags;
/* Will be called from irq-context when using foreign fences. */
spin_lock_irqsave(&engine->timeline->lock, flags);
spin_lock_irqsave(&engine->timeline.lock, flags);
queue_request(engine, &request->sched, rq_prio(request));
submit_queue(engine, rq_prio(request));
@ -1175,7 +1175,7 @@ static void execlists_submit_request(struct i915_request *request)
GEM_BUG_ON(!engine->execlists.first);
GEM_BUG_ON(list_empty(&request->sched.link));
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
}
static struct i915_request *sched_to_request(struct i915_sched_node *node)
@ -1191,8 +1191,8 @@ sched_lock_engine(struct i915_sched_node *node, struct intel_engine_cs *locked)
GEM_BUG_ON(!locked);
if (engine != locked) {
spin_unlock(&locked->timeline->lock);
spin_lock(&engine->timeline->lock);
spin_unlock(&locked->timeline.lock);
spin_lock(&engine->timeline.lock);
}
return engine;
@ -1275,7 +1275,7 @@ static void execlists_schedule(struct i915_request *request,
}
engine = request->engine;
spin_lock_irq(&engine->timeline->lock);
spin_lock_irq(&engine->timeline.lock);
/* Fifo and depth-first replacement ensure our deps execute before us */
list_for_each_entry_safe_reverse(dep, p, &dfs, dfs_link) {
@ -1299,7 +1299,7 @@ static void execlists_schedule(struct i915_request *request,
__submit_queue(engine, prio);
}
spin_unlock_irq(&engine->timeline->lock);
spin_unlock_irq(&engine->timeline.lock);
}
static int __context_pin(struct i915_gem_context *ctx, struct i915_vma *vma)
@ -1828,9 +1828,9 @@ static void reset_common_ring(struct intel_engine_cs *engine,
reset_irq(engine);
/* Push back any incomplete requests for replay after the reset. */
spin_lock(&engine->timeline->lock);
spin_lock(&engine->timeline.lock);
__unwind_incomplete_requests(engine);
spin_unlock(&engine->timeline->lock);
spin_unlock(&engine->timeline.lock);
local_irq_restore(flags);
@ -2599,6 +2599,7 @@ static int execlists_context_deferred_alloc(struct i915_gem_context *ctx,
struct i915_vma *vma;
uint32_t context_size;
struct intel_ring *ring;
struct i915_timeline *timeline;
int ret;
if (ce->state)
@ -2614,8 +2615,8 @@ static int execlists_context_deferred_alloc(struct i915_gem_context *ctx,
ctx_obj = i915_gem_object_create(ctx->i915, context_size);
if (IS_ERR(ctx_obj)) {
DRM_DEBUG_DRIVER("Alloc LRC backing obj failed.\n");
return PTR_ERR(ctx_obj);
ret = PTR_ERR(ctx_obj);
goto error_deref_obj;
}
vma = i915_vma_instance(ctx_obj, &ctx->i915->ggtt.base, NULL);
@ -2624,7 +2625,14 @@ static int execlists_context_deferred_alloc(struct i915_gem_context *ctx,
goto error_deref_obj;
}
ring = intel_engine_create_ring(engine, ctx->timeline, ctx->ring_size);
timeline = i915_timeline_create(ctx->i915, ctx->name);
if (IS_ERR(timeline)) {
ret = PTR_ERR(timeline);
goto error_deref_obj;
}
ring = intel_engine_create_ring(engine, timeline, ctx->ring_size);
i915_timeline_put(timeline);
if (IS_ERR(ring)) {
ret = PTR_ERR(ring);
goto error_deref_obj;

25
drivers/gpu/drm/i915/intel_ringbuffer.c
Просмотреть файл

@ -697,17 +697,17 @@ static void cancel_requests(struct intel_engine_cs *engine)
struct i915_request *request;
unsigned long flags;
spin_lock_irqsave(&engine->timeline->lock, flags);
spin_lock_irqsave(&engine->timeline.lock, flags);
/* Mark all submitted requests as skipped. */
list_for_each_entry(request, &engine->timeline->requests, link) {
list_for_each_entry(request, &engine->timeline.requests, link) {
GEM_BUG_ON(!request->global_seqno);
if (!i915_request_completed(request))
dma_fence_set_error(&request->fence, -EIO);
}
/* Remaining _unready_ requests will be nop'ed when submitted */
spin_unlock_irqrestore(&engine->timeline->lock, flags);
spin_unlock_irqrestore(&engine->timeline.lock, flags);
}
static void i9xx_submit_request(struct i915_request *request)
@ -1118,7 +1118,7 @@ err:
struct intel_ring *
intel_engine_create_ring(struct intel_engine_cs *engine,
struct i915_gem_timeline *timeline,
struct i915_timeline *timeline,
int size)
{
struct intel_ring *ring;
@ -1126,7 +1126,7 @@ intel_engine_create_ring(struct intel_engine_cs *engine,
GEM_BUG_ON(!is_power_of_2(size));
GEM_BUG_ON(RING_CTL_SIZE(size) & ~RING_NR_PAGES);
GEM_BUG_ON(&timeline->engine[engine->id] == engine->timeline);
GEM_BUG_ON(timeline == &engine->timeline);
lockdep_assert_held(&engine->i915->drm.struct_mutex);
ring = kzalloc(sizeof(*ring), GFP_KERNEL);
@ -1134,7 +1134,7 @@ intel_engine_create_ring(struct intel_engine_cs *engine,
return ERR_PTR(-ENOMEM);
INIT_LIST_HEAD(&ring->request_list);
ring->timeline = &timeline->engine[engine->id];
ring->timeline = i915_timeline_get(timeline);
ring->size = size;
/* Workaround an erratum on the i830 which causes a hang if
@ -1165,6 +1165,7 @@ intel_ring_free(struct intel_ring *ring)
i915_vma_close(ring->vma);
__i915_gem_object_release_unless_active(obj);
i915_timeline_put(ring->timeline);
kfree(ring);
}
@ -1323,6 +1324,7 @@ static void intel_ring_context_unpin(struct intel_engine_cs *engine,
static int intel_init_ring_buffer(struct intel_engine_cs *engine)
{
struct intel_ring *ring;
struct i915_timeline *timeline;
int err;
intel_engine_setup_common(engine);
@ -1331,9 +1333,14 @@ static int intel_init_ring_buffer(struct intel_engine_cs *engine)
if (err)
goto err;
ring = intel_engine_create_ring(engine,
&engine->i915->gt.legacy_timeline,
32 * PAGE_SIZE);
timeline = i915_timeline_create(engine->i915, engine->name);
if (IS_ERR(timeline)) {
err = PTR_ERR(timeline);
goto err;
}
ring = intel_engine_create_ring(engine, timeline, 32 * PAGE_SIZE);
i915_timeline_put(timeline);
if (IS_ERR(ring)) {
err = PTR_ERR(ring);
goto err;

11
drivers/gpu/drm/i915/intel_ringbuffer.h
Просмотреть файл

@ -6,12 +6,12 @@
#include <linux/seqlock.h>
#include "i915_gem_batch_pool.h"
#include "i915_gem_timeline.h"
#include "i915_reg.h"
#include "i915_pmu.h"
#include "i915_request.h"
#include "i915_selftest.h"
#include "i915_timeline.h"
#include "intel_gpu_commands.h"
struct drm_printer;
@ -129,7 +129,7 @@ struct intel_ring {
struct i915_vma *vma;
void *vaddr;
struct intel_timeline *timeline;
struct i915_timeline *timeline;
struct list_head request_list;
struct list_head active_link;
@ -338,7 +338,8 @@ struct intel_engine_cs {
u32 mmio_base;
struct intel_ring *buffer;
struct intel_timeline *timeline;
struct i915_timeline timeline;
struct drm_i915_gem_object *default_state;
@ -770,7 +771,7 @@ intel_write_status_page(struct intel_engine_cs *engine, int reg, u32 value)
struct intel_ring *
intel_engine_create_ring(struct intel_engine_cs *engine,
struct i915_gem_timeline *timeline,
struct i915_timeline *timeline,
int size);
int intel_ring_pin(struct intel_ring *ring,
struct drm_i915_private *i915,
@ -889,7 +890,7 @@ static inline u32 intel_engine_last_submit(struct intel_engine_cs *engine)
* wtih serialising this hint with anything, so document it as
* a hint and nothing more.
*/
return READ_ONCE(engine->timeline->seqno);
return READ_ONCE(engine->timeline.seqno);
}
void intel_engine_get_instdone(struct intel_engine_cs *engine,

12
drivers/gpu/drm/i915/selftests/i915_gem_context.c
Просмотреть файл

@ -355,18 +355,6 @@ static int igt_ctx_exec(void *arg)
if (first_shared_gtt) {
ctx = __create_hw_context(i915, file->driver_priv);
if (!IS_ERR(ctx) && HAS_EXECLISTS(i915)) {
struct i915_gem_timeline *timeline;
timeline = i915_gem_timeline_create(i915, ctx->name);
if (IS_ERR(timeline)) {
__destroy_hw_context(ctx, file->driver_priv);
ctx = ERR_CAST(timeline);
} else {
ctx->timeline = timeline;
}
}
first_shared_gtt = false;
} else {
ctx = i915_gem_create_context(i915, file->driver_priv);

94
drivers/gpu/drm/i915/selftests/i915_gem_timeline.c → drivers/gpu/drm/i915/selftests/i915_timeline.c
Просмотреть файл

@ -1,25 +1,7 @@
/*
* Copyright © 2017 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
* SPDX-License-Identifier: MIT
*
* Copyright © 2017-2018 Intel Corporation
*/
#include "../i915_selftest.h"
@ -35,21 +17,21 @@ struct __igt_sync {
bool set;
};
static int __igt_sync(struct intel_timeline *tl,
static int __igt_sync(struct i915_timeline *tl,
u64 ctx,
const struct __igt_sync *p,
const char *name)
{
int ret;
if (__intel_timeline_sync_is_later(tl, ctx, p->seqno) != p->expected) {
if (__i915_timeline_sync_is_later(tl, ctx, p->seqno) != p->expected) {
pr_err("%s: %s(ctx=%llu, seqno=%u) expected passed %s but failed\n",
name, p->name, ctx, p->seqno, yesno(p->expected));
return -EINVAL;
}
if (p->set) {
ret = __intel_timeline_sync_set(tl, ctx, p->seqno);
ret = __i915_timeline_sync_set(tl, ctx, p->seqno);
if (ret)
return ret;
}
@ -77,37 +59,31 @@ static int igt_sync(void *arg)
{ "unwrap", UINT_MAX, true, false },
{},
}, *p;
struct intel_timeline *tl;
struct i915_timeline tl;
int order, offset;
int ret = -ENODEV;
tl = mock_timeline(0);
if (!tl)
return -ENOMEM;
mock_timeline_init(&tl, 0);
for (p = pass; p->name; p++) {
for (order = 1; order < 64; order++) {
for (offset = -1; offset <= (order > 1); offset++) {
u64 ctx = BIT_ULL(order) + offset;
ret = __igt_sync(tl, ctx, p, "1");
ret = __igt_sync(&tl, ctx, p, "1");
if (ret)
goto out;
}
}
}
mock_timeline_destroy(tl);
tl = mock_timeline(0);
if (!tl)
return -ENOMEM;
mock_timeline_fini(&tl);
mock_timeline_init(&tl, 0);
for (order = 1; order < 64; order++) {
for (offset = -1; offset <= (order > 1); offset++) {
u64 ctx = BIT_ULL(order) + offset;
for (p = pass; p->name; p++) {
ret = __igt_sync(tl, ctx, p, "2");
ret = __igt_sync(&tl, ctx, p, "2");
if (ret)
goto out;
}
@ -115,7 +91,7 @@ static int igt_sync(void *arg)
}
out:
mock_timeline_destroy(tl);
mock_timeline_fini(&tl);
return ret;
}
@ -127,15 +103,13 @@ static unsigned int random_engine(struct rnd_state *rnd)
static int bench_sync(void *arg)
{
struct rnd_state prng;
struct intel_timeline *tl;
struct i915_timeline tl;
unsigned long end_time, count;
u64 prng32_1M;
ktime_t kt;
int order, last_order;
tl = mock_timeline(0);
if (!tl)
return -ENOMEM;
mock_timeline_init(&tl, 0);
/* Lookups from cache are very fast and so the random number generation
* and the loop itself becomes a significant factor in the per-iteration
@ -167,7 +141,7 @@ static int bench_sync(void *arg)
do {
u64 id = i915_prandom_u64_state(&prng);
__intel_timeline_sync_set(tl, id, 0);
__i915_timeline_sync_set(&tl, id, 0);
count++;
} while (!time_after(jiffies, end_time));
kt = ktime_sub(ktime_get(), kt);
@ -182,8 +156,8 @@ static int bench_sync(void *arg)
while (end_time--) {
u64 id = i915_prandom_u64_state(&prng);
if (!__intel_timeline_sync_is_later(tl, id, 0)) {
mock_timeline_destroy(tl);
if (!__i915_timeline_sync_is_later(&tl, id, 0)) {
mock_timeline_fini(&tl);
pr_err("Lookup of %llu failed\n", id);
return -EINVAL;
}
@ -193,19 +167,17 @@ static int bench_sync(void *arg)
pr_info("%s: %lu random lookups, %lluns/lookup\n",
__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));
mock_timeline_destroy(tl);
mock_timeline_fini(&tl);
cond_resched();
tl = mock_timeline(0);
if (!tl)
return -ENOMEM;
mock_timeline_init(&tl, 0);
/* Benchmark setting the first N (in order) contexts */
count = 0;
kt = ktime_get();
end_time = jiffies + HZ/10;
do {
__intel_timeline_sync_set(tl, count++, 0);
__i915_timeline_sync_set(&tl, count++, 0);
} while (!time_after(jiffies, end_time));
kt = ktime_sub(ktime_get(), kt);
pr_info("%s: %lu in-order insertions, %lluns/insert\n",
@ -215,9 +187,9 @@ static int bench_sync(void *arg)
end_time = count;
kt = ktime_get();
while (end_time--) {
if (!__intel_timeline_sync_is_later(tl, end_time, 0)) {
if (!__i915_timeline_sync_is_later(&tl, end_time, 0)) {
pr_err("Lookup of %lu failed\n", end_time);
mock_timeline_destroy(tl);
mock_timeline_fini(&tl);
return -EINVAL;
}
}
@ -225,12 +197,10 @@ static int bench_sync(void *arg)
pr_info("%s: %lu in-order lookups, %lluns/lookup\n",
__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));
mock_timeline_destroy(tl);
mock_timeline_fini(&tl);
cond_resched();
tl = mock_timeline(0);
if (!tl)
return -ENOMEM;
mock_timeline_init(&tl, 0);
/* Benchmark searching for a random context id and maybe changing it */
prandom_seed_state(&prng, i915_selftest.random_seed);
@ -241,8 +211,8 @@ static int bench_sync(void *arg)
u32 id = random_engine(&prng);
u32 seqno = prandom_u32_state(&prng);
if (!__intel_timeline_sync_is_later(tl, id, seqno))
__intel_timeline_sync_set(tl, id, seqno);
if (!__i915_timeline_sync_is_later(&tl, id, seqno))
__i915_timeline_sync_set(&tl, id, seqno);
count++;
} while (!time_after(jiffies, end_time));
@ -250,7 +220,7 @@ static int bench_sync(void *arg)
kt = ktime_sub_ns(kt, (count * prng32_1M * 2) >> 20);
pr_info("%s: %lu repeated insert/lookups, %lluns/op\n",
__func__, count, (long long)div64_ul(ktime_to_ns(kt), count));
mock_timeline_destroy(tl);
mock_timeline_fini(&tl);
cond_resched();
/* Benchmark searching for a known context id and changing the seqno */
@ -258,9 +228,7 @@ static int bench_sync(void *arg)
({ int tmp = last_order; last_order = order; order += tmp; })) {
unsigned int mask = BIT(order) - 1;
tl = mock_timeline(0);
if (!tl)
return -ENOMEM;
mock_timeline_init(&tl, 0);
count = 0;
kt = ktime_get();
@ -272,8 +240,8 @@ static int bench_sync(void *arg)
*/
u64 id = (u64)(count & mask) << order;
__intel_timeline_sync_is_later(tl, id, 0);
__intel_timeline_sync_set(tl, id, 0);
__i915_timeline_sync_is_later(&tl, id, 0);
__i915_timeline_sync_set(&tl, id, 0);
count++;
} while (!time_after(jiffies, end_time));
@ -281,7 +249,7 @@ static int bench_sync(void *arg)
pr_info("%s: %lu cyclic/%d insert/lookups, %lluns/op\n",
__func__, count, order,
(long long)div64_ul(ktime_to_ns(kt), count));
mock_timeline_destroy(tl);
mock_timeline_fini(&tl);
cond_resched();
}

32
drivers/gpu/drm/i915/selftests/mock_engine.c
Просмотреть файл

@ -25,6 +25,11 @@
#include "mock_engine.h"
#include "mock_request.h"
struct mock_ring {
struct intel_ring base;
struct i915_timeline timeline;
};
static struct mock_request *first_request(struct mock_engine *engine)
{
return list_first_entry_or_null(&engine->hw_queue,
@ -132,7 +137,7 @@ static void mock_submit_request(struct i915_request *request)
static struct intel_ring *mock_ring(struct intel_engine_cs *engine)
{
const unsigned long sz = PAGE_SIZE / 2;
struct intel_ring *ring;
struct mock_ring *ring;
BUILD_BUG_ON(MIN_SPACE_FOR_ADD_REQUEST > sz);
@ -140,20 +145,24 @@ static struct intel_ring *mock_ring(struct intel_engine_cs *engine)
if (!ring)
return NULL;
ring->timeline = &engine->i915->gt.legacy_timeline.engine[engine->id];
i915_timeline_init(engine->i915, &ring->timeline, engine->name);
ring->size = sz;
ring->effective_size = sz;
ring->vaddr = (void *)(ring + 1);
ring->base.size = sz;
ring->base.effective_size = sz;
ring->base.vaddr = (void *)(ring + 1);
ring->base.timeline = &ring->timeline;
INIT_LIST_HEAD(&ring->request_list);
intel_ring_update_space(ring);
INIT_LIST_HEAD(&ring->base.request_list);
intel_ring_update_space(&ring->base);
return ring;
return &ring->base;
}
static void mock_ring_free(struct intel_ring *ring)
static void mock_ring_free(struct intel_ring *base)
{
struct mock_ring *ring = container_of(base, typeof(*ring), base);
i915_timeline_fini(&ring->timeline);
kfree(ring);
}
@ -182,8 +191,7 @@ struct intel_engine_cs *mock_engine(struct drm_i915_private *i915,
engine->base.emit_breadcrumb = mock_emit_breadcrumb;
engine->base.submit_request = mock_submit_request;
intel_engine_init_timeline(&engine->base);
i915_timeline_init(i915, &engine->base.timeline, engine->base.name);
intel_engine_init_breadcrumbs(&engine->base);
engine->base.breadcrumbs.mock = true; /* prevent touching HW for irqs */
@ -200,6 +208,7 @@ struct intel_engine_cs *mock_engine(struct drm_i915_private *i915,
err_breadcrumbs:
intel_engine_fini_breadcrumbs(&engine->base);
i915_timeline_fini(&engine->base.timeline);
kfree(engine);
return NULL;
}
@ -238,6 +247,7 @@ void mock_engine_free(struct intel_engine_cs *engine)
mock_ring_free(engine->buffer);
intel_engine_fini_breadcrumbs(engine);
i915_timeline_fini(&engine->timeline);
kfree(engine);
}

10
drivers/gpu/drm/i915/selftests/mock_gem_device.c
Просмотреть файл

@ -73,10 +73,8 @@ static void mock_device_release(struct drm_device *dev)
mutex_lock(&i915->drm.struct_mutex);
mock_fini_ggtt(i915);
i915_gem_timeline_fini(&i915->gt.legacy_timeline);
i915_gem_timeline_fini(&i915->gt.execution_timeline);
WARN_ON(!list_empty(&i915->gt.timelines));
mutex_unlock(&i915->drm.struct_mutex);
WARN_ON(!list_empty(&i915->gt.timelines));
destroy_workqueue(i915->wq);
@ -230,12 +228,6 @@ struct drm_i915_private *mock_gem_device(void)
INIT_LIST_HEAD(&i915->gt.active_rings);
mutex_lock(&i915->drm.struct_mutex);
err = i915_gem_timeline_init__global(i915);
if (err) {
mutex_unlock(&i915->drm.struct_mutex);
goto err_priorities;
}
mock_init_ggtt(i915);
mutex_unlock(&i915->drm.struct_mutex);

45
drivers/gpu/drm/i915/selftests/mock_timeline.c
Просмотреть файл

@ -1,45 +1,28 @@
/*
* Copyright © 2017 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
* SPDX-License-Identifier: MIT
*
* Copyright © 2017-2018 Intel Corporation
*/
#include "../i915_timeline.h"
#include "mock_timeline.h"
struct intel_timeline *mock_timeline(u64 context)
void mock_timeline_init(struct i915_timeline *timeline, u64 context)
{
static struct lock_class_key class;
struct intel_timeline *tl;
timeline->fence_context = context;
tl = kzalloc(sizeof(*tl), GFP_KERNEL);
if (!tl)
return NULL;
spin_lock_init(&timeline->lock);
__intel_timeline_init(tl, NULL, context, &class, "mock");
init_request_active(&timeline->last_request, NULL);
INIT_LIST_HEAD(&timeline->requests);
return tl;
i915_syncmap_init(&timeline->sync);
INIT_LIST_HEAD(&timeline->link);
}
void mock_timeline_destroy(struct intel_timeline *tl)
void mock_timeline_fini(struct i915_timeline *timeline)
{
__intel_timeline_fini(tl);
kfree(tl);
i915_timeline_fini(timeline);
}

28
drivers/gpu/drm/i915/selftests/mock_timeline.h
Просмотреть файл

@ -1,33 +1,15 @@
/*
* Copyright © 2017 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
* SPDX-License-Identifier: MIT
*
* Copyright © 2017-2018 Intel Corporation
*/
#ifndef __MOCK_TIMELINE__
#define __MOCK_TIMELINE__
#include "../i915_gem_timeline.h"
struct i915_timeline;
struct intel_timeline *mock_timeline(u64 context);
void mock_timeline_destroy(struct intel_timeline *tl);
void mock_timeline_init(struct i915_timeline *timeline, u64 context);
void mock_timeline_fini(struct i915_timeline *timeline);
#endif /* !__MOCK_TIMELINE__ */