mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-23 08:56:59 +07:00
54ff01fd0d
This trys to give new born vGPU with higher scheduling chance not only with adding to sched list head and also have higher priority for workload sched for 2 seconds after starting to schedule it. In order for fast GPU execution during VM boot, and ensure guest driver setup with required state given in time. This fixes recent failure seen on one VM with multiple linux VMs running on kernel with commit 2621cefaa42b3("drm/i915: Provide a timeout to i915_gem_wait_for_idle() on setup"), which had shorter setup timeout that caused context state init failed. v2: change to 2s for higher scheduling period Cc: Yuan Hang <hang.yuan@intel.com> Reviewed-by: Hang Yuan <hang.yuan@intel.com> Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
480 lines
12 KiB
C
480 lines
12 KiB
C
/*
|
|
* Copyright(c) 2011-2016 Intel Corporation. All rights reserved.
|
|
*
|
|
* 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.
|
|
*
|
|
* Authors:
|
|
* Anhua Xu
|
|
* Kevin Tian <kevin.tian@intel.com>
|
|
*
|
|
* Contributors:
|
|
* Min He <min.he@intel.com>
|
|
* Bing Niu <bing.niu@intel.com>
|
|
* Zhi Wang <zhi.a.wang@intel.com>
|
|
*
|
|
*/
|
|
|
|
#include "i915_drv.h"
|
|
#include "gvt.h"
|
|
|
|
static bool vgpu_has_pending_workload(struct intel_vgpu *vgpu)
|
|
{
|
|
enum intel_engine_id i;
|
|
struct intel_engine_cs *engine;
|
|
|
|
for_each_engine(engine, vgpu->gvt->dev_priv, i) {
|
|
if (!list_empty(workload_q_head(vgpu, i)))
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
/* We give 2 seconds higher prio for vGPU during start */
|
|
#define GVT_SCHED_VGPU_PRI_TIME 2
|
|
|
|
struct vgpu_sched_data {
|
|
struct list_head lru_list;
|
|
struct intel_vgpu *vgpu;
|
|
bool active;
|
|
bool pri_sched;
|
|
ktime_t pri_time;
|
|
ktime_t sched_in_time;
|
|
ktime_t sched_time;
|
|
ktime_t left_ts;
|
|
ktime_t allocated_ts;
|
|
|
|
struct vgpu_sched_ctl sched_ctl;
|
|
};
|
|
|
|
struct gvt_sched_data {
|
|
struct intel_gvt *gvt;
|
|
struct hrtimer timer;
|
|
unsigned long period;
|
|
struct list_head lru_runq_head;
|
|
ktime_t expire_time;
|
|
};
|
|
|
|
static void vgpu_update_timeslice(struct intel_vgpu *vgpu, ktime_t cur_time)
|
|
{
|
|
ktime_t delta_ts;
|
|
struct vgpu_sched_data *vgpu_data;
|
|
|
|
if (!vgpu || vgpu == vgpu->gvt->idle_vgpu)
|
|
return;
|
|
|
|
vgpu_data = vgpu->sched_data;
|
|
delta_ts = ktime_sub(cur_time, vgpu_data->sched_in_time);
|
|
vgpu_data->sched_time = ktime_add(vgpu_data->sched_time, delta_ts);
|
|
vgpu_data->left_ts = ktime_sub(vgpu_data->left_ts, delta_ts);
|
|
vgpu_data->sched_in_time = cur_time;
|
|
}
|
|
|
|
#define GVT_TS_BALANCE_PERIOD_MS 100
|
|
#define GVT_TS_BALANCE_STAGE_NUM 10
|
|
|
|
static void gvt_balance_timeslice(struct gvt_sched_data *sched_data)
|
|
{
|
|
struct vgpu_sched_data *vgpu_data;
|
|
struct list_head *pos;
|
|
static uint64_t stage_check;
|
|
int stage = stage_check++ % GVT_TS_BALANCE_STAGE_NUM;
|
|
|
|
/* The timeslice accumulation reset at stage 0, which is
|
|
* allocated again without adding previous debt.
|
|
*/
|
|
if (stage == 0) {
|
|
int total_weight = 0;
|
|
ktime_t fair_timeslice;
|
|
|
|
list_for_each(pos, &sched_data->lru_runq_head) {
|
|
vgpu_data = container_of(pos, struct vgpu_sched_data, lru_list);
|
|
total_weight += vgpu_data->sched_ctl.weight;
|
|
}
|
|
|
|
list_for_each(pos, &sched_data->lru_runq_head) {
|
|
vgpu_data = container_of(pos, struct vgpu_sched_data, lru_list);
|
|
fair_timeslice = ktime_divns(ms_to_ktime(GVT_TS_BALANCE_PERIOD_MS),
|
|
total_weight) * vgpu_data->sched_ctl.weight;
|
|
|
|
vgpu_data->allocated_ts = fair_timeslice;
|
|
vgpu_data->left_ts = vgpu_data->allocated_ts;
|
|
}
|
|
} else {
|
|
list_for_each(pos, &sched_data->lru_runq_head) {
|
|
vgpu_data = container_of(pos, struct vgpu_sched_data, lru_list);
|
|
|
|
/* timeslice for next 100ms should add the left/debt
|
|
* slice of previous stages.
|
|
*/
|
|
vgpu_data->left_ts += vgpu_data->allocated_ts;
|
|
}
|
|
}
|
|
}
|
|
|
|
static void try_to_schedule_next_vgpu(struct intel_gvt *gvt)
|
|
{
|
|
struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
|
|
enum intel_engine_id i;
|
|
struct intel_engine_cs *engine;
|
|
struct vgpu_sched_data *vgpu_data;
|
|
ktime_t cur_time;
|
|
|
|
/* no need to schedule if next_vgpu is the same with current_vgpu,
|
|
* let scheduler chose next_vgpu again by setting it to NULL.
|
|
*/
|
|
if (scheduler->next_vgpu == scheduler->current_vgpu) {
|
|
scheduler->next_vgpu = NULL;
|
|
return;
|
|
}
|
|
|
|
/*
|
|
* after the flag is set, workload dispatch thread will
|
|
* stop dispatching workload for current vgpu
|
|
*/
|
|
scheduler->need_reschedule = true;
|
|
|
|
/* still have uncompleted workload? */
|
|
for_each_engine(engine, gvt->dev_priv, i) {
|
|
if (scheduler->current_workload[i])
|
|
return;
|
|
}
|
|
|
|
cur_time = ktime_get();
|
|
vgpu_update_timeslice(scheduler->current_vgpu, cur_time);
|
|
vgpu_data = scheduler->next_vgpu->sched_data;
|
|
vgpu_data->sched_in_time = cur_time;
|
|
|
|
/* switch current vgpu */
|
|
scheduler->current_vgpu = scheduler->next_vgpu;
|
|
scheduler->next_vgpu = NULL;
|
|
|
|
scheduler->need_reschedule = false;
|
|
|
|
/* wake up workload dispatch thread */
|
|
for_each_engine(engine, gvt->dev_priv, i)
|
|
wake_up(&scheduler->waitq[i]);
|
|
}
|
|
|
|
static struct intel_vgpu *find_busy_vgpu(struct gvt_sched_data *sched_data)
|
|
{
|
|
struct vgpu_sched_data *vgpu_data;
|
|
struct intel_vgpu *vgpu = NULL;
|
|
struct list_head *head = &sched_data->lru_runq_head;
|
|
struct list_head *pos;
|
|
|
|
/* search a vgpu with pending workload */
|
|
list_for_each(pos, head) {
|
|
|
|
vgpu_data = container_of(pos, struct vgpu_sched_data, lru_list);
|
|
if (!vgpu_has_pending_workload(vgpu_data->vgpu))
|
|
continue;
|
|
|
|
if (vgpu_data->pri_sched) {
|
|
if (ktime_before(ktime_get(), vgpu_data->pri_time)) {
|
|
vgpu = vgpu_data->vgpu;
|
|
break;
|
|
} else
|
|
vgpu_data->pri_sched = false;
|
|
}
|
|
|
|
/* Return the vGPU only if it has time slice left */
|
|
if (vgpu_data->left_ts > 0) {
|
|
vgpu = vgpu_data->vgpu;
|
|
break;
|
|
}
|
|
}
|
|
|
|
return vgpu;
|
|
}
|
|
|
|
/* in nanosecond */
|
|
#define GVT_DEFAULT_TIME_SLICE 1000000
|
|
|
|
static void tbs_sched_func(struct gvt_sched_data *sched_data)
|
|
{
|
|
struct intel_gvt *gvt = sched_data->gvt;
|
|
struct intel_gvt_workload_scheduler *scheduler = &gvt->scheduler;
|
|
struct vgpu_sched_data *vgpu_data;
|
|
struct intel_vgpu *vgpu = NULL;
|
|
|
|
/* no active vgpu or has already had a target */
|
|
if (list_empty(&sched_data->lru_runq_head) || scheduler->next_vgpu)
|
|
goto out;
|
|
|
|
vgpu = find_busy_vgpu(sched_data);
|
|
if (vgpu) {
|
|
scheduler->next_vgpu = vgpu;
|
|
vgpu_data = vgpu->sched_data;
|
|
if (!vgpu_data->pri_sched) {
|
|
/* Move the last used vGPU to the tail of lru_list */
|
|
list_del_init(&vgpu_data->lru_list);
|
|
list_add_tail(&vgpu_data->lru_list,
|
|
&sched_data->lru_runq_head);
|
|
}
|
|
} else {
|
|
scheduler->next_vgpu = gvt->idle_vgpu;
|
|
}
|
|
out:
|
|
if (scheduler->next_vgpu)
|
|
try_to_schedule_next_vgpu(gvt);
|
|
}
|
|
|
|
void intel_gvt_schedule(struct intel_gvt *gvt)
|
|
{
|
|
struct gvt_sched_data *sched_data = gvt->scheduler.sched_data;
|
|
ktime_t cur_time;
|
|
|
|
mutex_lock(&gvt->sched_lock);
|
|
cur_time = ktime_get();
|
|
|
|
if (test_and_clear_bit(INTEL_GVT_REQUEST_SCHED,
|
|
(void *)&gvt->service_request)) {
|
|
if (cur_time >= sched_data->expire_time) {
|
|
gvt_balance_timeslice(sched_data);
|
|
sched_data->expire_time = ktime_add_ms(
|
|
cur_time, GVT_TS_BALANCE_PERIOD_MS);
|
|
}
|
|
}
|
|
clear_bit(INTEL_GVT_REQUEST_EVENT_SCHED, (void *)&gvt->service_request);
|
|
|
|
vgpu_update_timeslice(gvt->scheduler.current_vgpu, cur_time);
|
|
tbs_sched_func(sched_data);
|
|
|
|
mutex_unlock(&gvt->sched_lock);
|
|
}
|
|
|
|
static enum hrtimer_restart tbs_timer_fn(struct hrtimer *timer_data)
|
|
{
|
|
struct gvt_sched_data *data;
|
|
|
|
data = container_of(timer_data, struct gvt_sched_data, timer);
|
|
|
|
intel_gvt_request_service(data->gvt, INTEL_GVT_REQUEST_SCHED);
|
|
|
|
hrtimer_add_expires_ns(&data->timer, data->period);
|
|
|
|
return HRTIMER_RESTART;
|
|
}
|
|
|
|
static int tbs_sched_init(struct intel_gvt *gvt)
|
|
{
|
|
struct intel_gvt_workload_scheduler *scheduler =
|
|
&gvt->scheduler;
|
|
|
|
struct gvt_sched_data *data;
|
|
|
|
data = kzalloc(sizeof(*data), GFP_KERNEL);
|
|
if (!data)
|
|
return -ENOMEM;
|
|
|
|
INIT_LIST_HEAD(&data->lru_runq_head);
|
|
hrtimer_init(&data->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
|
|
data->timer.function = tbs_timer_fn;
|
|
data->period = GVT_DEFAULT_TIME_SLICE;
|
|
data->gvt = gvt;
|
|
|
|
scheduler->sched_data = data;
|
|
|
|
return 0;
|
|
}
|
|
|
|
static void tbs_sched_clean(struct intel_gvt *gvt)
|
|
{
|
|
struct intel_gvt_workload_scheduler *scheduler =
|
|
&gvt->scheduler;
|
|
struct gvt_sched_data *data = scheduler->sched_data;
|
|
|
|
hrtimer_cancel(&data->timer);
|
|
|
|
kfree(data);
|
|
scheduler->sched_data = NULL;
|
|
}
|
|
|
|
static int tbs_sched_init_vgpu(struct intel_vgpu *vgpu)
|
|
{
|
|
struct vgpu_sched_data *data;
|
|
|
|
data = kzalloc(sizeof(*data), GFP_KERNEL);
|
|
if (!data)
|
|
return -ENOMEM;
|
|
|
|
data->sched_ctl.weight = vgpu->sched_ctl.weight;
|
|
data->vgpu = vgpu;
|
|
INIT_LIST_HEAD(&data->lru_list);
|
|
|
|
vgpu->sched_data = data;
|
|
|
|
return 0;
|
|
}
|
|
|
|
static void tbs_sched_clean_vgpu(struct intel_vgpu *vgpu)
|
|
{
|
|
struct intel_gvt *gvt = vgpu->gvt;
|
|
struct gvt_sched_data *sched_data = gvt->scheduler.sched_data;
|
|
|
|
kfree(vgpu->sched_data);
|
|
vgpu->sched_data = NULL;
|
|
|
|
/* this vgpu id has been removed */
|
|
if (idr_is_empty(&gvt->vgpu_idr))
|
|
hrtimer_cancel(&sched_data->timer);
|
|
}
|
|
|
|
static void tbs_sched_start_schedule(struct intel_vgpu *vgpu)
|
|
{
|
|
struct gvt_sched_data *sched_data = vgpu->gvt->scheduler.sched_data;
|
|
struct vgpu_sched_data *vgpu_data = vgpu->sched_data;
|
|
ktime_t now;
|
|
|
|
if (!list_empty(&vgpu_data->lru_list))
|
|
return;
|
|
|
|
now = ktime_get();
|
|
vgpu_data->pri_time = ktime_add(now,
|
|
ktime_set(GVT_SCHED_VGPU_PRI_TIME, 0));
|
|
vgpu_data->pri_sched = true;
|
|
|
|
list_add(&vgpu_data->lru_list, &sched_data->lru_runq_head);
|
|
|
|
if (!hrtimer_active(&sched_data->timer))
|
|
hrtimer_start(&sched_data->timer, ktime_add_ns(ktime_get(),
|
|
sched_data->period), HRTIMER_MODE_ABS);
|
|
vgpu_data->active = true;
|
|
}
|
|
|
|
static void tbs_sched_stop_schedule(struct intel_vgpu *vgpu)
|
|
{
|
|
struct vgpu_sched_data *vgpu_data = vgpu->sched_data;
|
|
|
|
list_del_init(&vgpu_data->lru_list);
|
|
vgpu_data->active = false;
|
|
}
|
|
|
|
static struct intel_gvt_sched_policy_ops tbs_schedule_ops = {
|
|
.init = tbs_sched_init,
|
|
.clean = tbs_sched_clean,
|
|
.init_vgpu = tbs_sched_init_vgpu,
|
|
.clean_vgpu = tbs_sched_clean_vgpu,
|
|
.start_schedule = tbs_sched_start_schedule,
|
|
.stop_schedule = tbs_sched_stop_schedule,
|
|
};
|
|
|
|
int intel_gvt_init_sched_policy(struct intel_gvt *gvt)
|
|
{
|
|
int ret;
|
|
|
|
mutex_lock(&gvt->sched_lock);
|
|
gvt->scheduler.sched_ops = &tbs_schedule_ops;
|
|
ret = gvt->scheduler.sched_ops->init(gvt);
|
|
mutex_unlock(&gvt->sched_lock);
|
|
|
|
return ret;
|
|
}
|
|
|
|
void intel_gvt_clean_sched_policy(struct intel_gvt *gvt)
|
|
{
|
|
mutex_lock(&gvt->sched_lock);
|
|
gvt->scheduler.sched_ops->clean(gvt);
|
|
mutex_unlock(&gvt->sched_lock);
|
|
}
|
|
|
|
/* for per-vgpu scheduler policy, there are 2 per-vgpu data:
|
|
* sched_data, and sched_ctl. We see these 2 data as part of
|
|
* the global scheduler which are proteced by gvt->sched_lock.
|
|
* Caller should make their decision if the vgpu_lock should
|
|
* be hold outside.
|
|
*/
|
|
|
|
int intel_vgpu_init_sched_policy(struct intel_vgpu *vgpu)
|
|
{
|
|
int ret;
|
|
|
|
mutex_lock(&vgpu->gvt->sched_lock);
|
|
ret = vgpu->gvt->scheduler.sched_ops->init_vgpu(vgpu);
|
|
mutex_unlock(&vgpu->gvt->sched_lock);
|
|
|
|
return ret;
|
|
}
|
|
|
|
void intel_vgpu_clean_sched_policy(struct intel_vgpu *vgpu)
|
|
{
|
|
mutex_lock(&vgpu->gvt->sched_lock);
|
|
vgpu->gvt->scheduler.sched_ops->clean_vgpu(vgpu);
|
|
mutex_unlock(&vgpu->gvt->sched_lock);
|
|
}
|
|
|
|
void intel_vgpu_start_schedule(struct intel_vgpu *vgpu)
|
|
{
|
|
struct vgpu_sched_data *vgpu_data = vgpu->sched_data;
|
|
|
|
mutex_lock(&vgpu->gvt->sched_lock);
|
|
if (!vgpu_data->active) {
|
|
gvt_dbg_core("vgpu%d: start schedule\n", vgpu->id);
|
|
vgpu->gvt->scheduler.sched_ops->start_schedule(vgpu);
|
|
}
|
|
mutex_unlock(&vgpu->gvt->sched_lock);
|
|
}
|
|
|
|
void intel_gvt_kick_schedule(struct intel_gvt *gvt)
|
|
{
|
|
mutex_lock(&gvt->sched_lock);
|
|
intel_gvt_request_service(gvt, INTEL_GVT_REQUEST_EVENT_SCHED);
|
|
mutex_unlock(&gvt->sched_lock);
|
|
}
|
|
|
|
void intel_vgpu_stop_schedule(struct intel_vgpu *vgpu)
|
|
{
|
|
struct intel_gvt_workload_scheduler *scheduler =
|
|
&vgpu->gvt->scheduler;
|
|
int ring_id;
|
|
struct vgpu_sched_data *vgpu_data = vgpu->sched_data;
|
|
struct drm_i915_private *dev_priv = vgpu->gvt->dev_priv;
|
|
|
|
if (!vgpu_data->active)
|
|
return;
|
|
|
|
gvt_dbg_core("vgpu%d: stop schedule\n", vgpu->id);
|
|
|
|
mutex_lock(&vgpu->gvt->sched_lock);
|
|
scheduler->sched_ops->stop_schedule(vgpu);
|
|
|
|
if (scheduler->next_vgpu == vgpu)
|
|
scheduler->next_vgpu = NULL;
|
|
|
|
if (scheduler->current_vgpu == vgpu) {
|
|
/* stop workload dispatching */
|
|
scheduler->need_reschedule = true;
|
|
scheduler->current_vgpu = NULL;
|
|
}
|
|
|
|
intel_runtime_pm_get(dev_priv);
|
|
spin_lock_bh(&scheduler->mmio_context_lock);
|
|
for (ring_id = 0; ring_id < I915_NUM_ENGINES; ring_id++) {
|
|
if (scheduler->engine_owner[ring_id] == vgpu) {
|
|
intel_gvt_switch_mmio(vgpu, NULL, ring_id);
|
|
scheduler->engine_owner[ring_id] = NULL;
|
|
}
|
|
}
|
|
spin_unlock_bh(&scheduler->mmio_context_lock);
|
|
intel_runtime_pm_put(dev_priv);
|
|
mutex_unlock(&vgpu->gvt->sched_lock);
|
|
}
|