提交 b92326a0 编写于 作者: C Chris Wilson

drm/i915: Only report a wakeup if the waiter was truly asleep

If we attempt to wake up a waiter, who is currently checking the seqno
it will be in the TASK_INTERRUPTIBLE state and ttwu will report success.
However, it is actually awake and functioning -- so delay reporting the
actual wake up until it sleeps. This fixes some spurious claims of
missed_breadcrumbs when running under heavy load; i.e. sufficient load to
preempt away the newly woken waiter before they complete their checks.
However, it does so at the cost of a rare false negative; where the
waiter changes between the check and ttwu -- the only way to fix that
would be to extend the reporting from ttwu where the check could be done
atomically.

v2: Defend against !CONFIG_SMP
v3: Don't filter out calls to wake_up_process
v4: Drop risky microoptimisation to skip wakeups

Testcase: igt/drv_missed_irq # sanity check we do detect missed_breadcrumb()
Testcase: igt/gem_concurrent_blit # for generating false positives
References: https://bugs.freedesktop.org/show_bug.cgi?id=100007Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171209124710.1606-1-chris@chris-wilson.co.ukReviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
上级 776bc27f
......@@ -27,6 +27,12 @@
#include "i915_drv.h"
#ifdef CONFIG_SMP
#define task_asleep(tsk) ((tsk)->state & TASK_NORMAL && !(tsk)->on_cpu)
#else
#define task_asleep(tsk) ((tsk)->state & TASK_NORMAL)
#endif
static unsigned int __intel_breadcrumbs_wakeup(struct intel_breadcrumbs *b)
{
struct intel_wait *wait;
......@@ -36,8 +42,20 @@ static unsigned int __intel_breadcrumbs_wakeup(struct intel_breadcrumbs *b)
wait = b->irq_wait;
if (wait) {
/*
* N.B. Since task_asleep() and ttwu are not atomic, the
* waiter may actually go to sleep after the check, causing
* us to suppress a valid wakeup. We prefer to reduce the
* number of false positive missed_breadcrumb() warnings
* at the expense of a few false negatives, as it it easy
* to trigger a false positive under heavy load. Enough
* signal should remain from genuine missed_breadcrumb()
* for us to detect in CI.
*/
bool was_asleep = task_asleep(wait->tsk);
result = ENGINE_WAKEUP_WAITER;
if (wake_up_process(wait->tsk))
if (wake_up_process(wait->tsk) && was_asleep)
result |= ENGINE_WAKEUP_ASLEEP;
}
......@@ -77,8 +95,8 @@ static noinline void missed_breadcrumb(struct intel_engine_cs *engine)
static void intel_breadcrumbs_hangcheck(struct timer_list *t)
{
struct intel_engine_cs *engine = from_timer(engine, t,
breadcrumbs.hangcheck);
struct intel_engine_cs *engine =
from_timer(engine, t, breadcrumbs.hangcheck);
struct intel_breadcrumbs *b = &engine->breadcrumbs;
if (!b->irq_armed)
......@@ -104,7 +122,7 @@ static void intel_breadcrumbs_hangcheck(struct timer_list *t)
*/
if (intel_engine_wakeup(engine) & ENGINE_WAKEUP_ASLEEP) {
missed_breadcrumb(engine);
mod_timer(&engine->breadcrumbs.fake_irq, jiffies + 1);
mod_timer(&b->fake_irq, jiffies + 1);
} else {
mod_timer(&b->hangcheck, wait_timeout());
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册