[PREEMPT_RT PATCH 0/3] i915 fixups for lockdep/lockdebugging

linux-rt-users.vger.kernel.org archive mirror
 help / color / mirror / Atom feed

* [PREEMPT_RT PATCH 0/3] i915 fixups for lockdep/lockdebugging
@ 2019-08-20  0:33 Clark Williams
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT Clark Williams
                   ` (2 more replies)
  0 siblings, 3 replies; 10+ messages in thread
From: Clark Williams @ 2019-08-20  0:33 UTC (permalink / raw)
  To: bigeasy; +Cc: tglx, linux-rt-users, linux-kernel

From: Clark Williams <williams@redhat.com>

The i915 driver was throwing splats on my home test box running
v5.2-rt3 when I turned on lockdep and lock debugging configs. This was 
mainly due to the non-side effects of the spin*_irq*() macros which do
nothing to IRQs on PREEMPT_RT. Converting the various irq_lock spinlocks
and the uncore lock to be raw_spinlock_t fixes this, although I'm not
sure of the performance implications for latency spikes. 

Testing was done on an Intel NUC with the following graphics:

00:02.0 VGA compatible controller: Intel Corporation Iris Plus Graphics 650 (rev 06)

Testing on other variants of i915 would be appreciated.

Clark Williams (3):
  i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT
  i915: convert all irq_locks spinlocks to raw spinlocks
  i915: convert uncore lock to raw spinlock

 drivers/gpu/drm/i915/i915_debugfs.c        |   8 +-
 drivers/gpu/drm/i915/i915_drv.c            |   6 +-
 drivers/gpu/drm/i915/i915_drv.h            |   2 +-
 drivers/gpu/drm/i915/i915_gem.c            |   4 +-
 drivers/gpu/drm/i915/i915_irq.c            | 154 ++++++++++-----------
 drivers/gpu/drm/i915/i915_pmu.c            |   4 +-
 drivers/gpu/drm/i915/intel_breadcrumbs.c   |  51 +++----
 drivers/gpu/drm/i915/intel_display.c       |  20 +--
 drivers/gpu/drm/i915/intel_engine_cs.c     |   4 +-
 drivers/gpu/drm/i915/intel_engine_types.h  |   2 +-
 drivers/gpu/drm/i915/intel_fifo_underrun.c |  16 +--
 drivers/gpu/drm/i915/intel_guc.c           |   6 +-
 drivers/gpu/drm/i915/intel_guc.h           |  10 +-
 drivers/gpu/drm/i915/intel_hotplug.c       |  36 ++---
 drivers/gpu/drm/i915/intel_pm.c            |   8 +-
 drivers/gpu/drm/i915/intel_runtime_pm.c    |   8 +-
 drivers/gpu/drm/i915/intel_sprite.c        |  32 ++---
 drivers/gpu/drm/i915/intel_tv.c            |   8 +-
 drivers/gpu/drm/i915/intel_uncore.c        |  52 +++----
 drivers/gpu/drm/i915/intel_uncore.h        |   2 +-
 drivers/gpu/drm/i915/intel_workarounds.c   |   4 +-
 21 files changed, 220 insertions(+), 217 deletions(-)

-- 
2.21.0


^ permalink raw reply	[flat|nested] 10+ messages in thread

* [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT
  2019-08-20  0:33 [PREEMPT_RT PATCH 0/3] i915 fixups for lockdep/lockdebugging Clark Williams
@ 2019-08-20  0:33 ` Clark Williams
  2019-09-25 16:47   ` Sebastian Andrzej Siewior
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks Clark Williams
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 3/3] i915: convert uncore lock to raw spinlock Clark Williams
  2 siblings, 1 reply; 10+ messages in thread
From: Clark Williams @ 2019-08-20  0:33 UTC (permalink / raw)
  To: bigeasy; +Cc: tglx, linux-rt-users, linux-kernel

From: Clark Williams <williams@redhat.com>

The 'breadcrumb' code in the i915 driver calls lockdep_assert_irqs_disabled()
when starting some operations. This is valid on a stock kernel
but on a PREEMPT_RT kernel the spin_lock_irq*() calls to not disable
interrupts and likewise the spin_unlock_irq*() calls to not enable interrupts.

Conditionalize these calls based on whether PREEMPT_RT_FULL is enabled.

Signed-off-by: Clark Williams <williams@redhat.com>
---
 drivers/gpu/drm/i915/intel_breadcrumbs.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
index 832cb6b1e9bd..3eef6010ebf6 100644
--- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
+++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
@@ -101,7 +101,8 @@ __dma_fence_signal__notify(struct dma_fence *fence)
 	struct dma_fence_cb *cur, *tmp;
 
 	lockdep_assert_held(fence->lock);
-	lockdep_assert_irqs_disabled();
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT_FULL))
+		lockdep_assert_irqs_disabled();
 
 	list_for_each_entry_safe(cur, tmp, &fence->cb_list, node) {
 		INIT_LIST_HEAD(&cur->node);
@@ -276,7 +277,8 @@ void intel_engine_fini_breadcrumbs(struct intel_engine_cs *engine)
 bool i915_request_enable_breadcrumb(struct i915_request *rq)
 {
 	lockdep_assert_held(&rq->lock);
-	lockdep_assert_irqs_disabled();
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT_FULL))
+		lockdep_assert_irqs_disabled();
 
 	if (test_bit(I915_FENCE_FLAG_ACTIVE, &rq->fence.flags)) {
 		struct intel_breadcrumbs *b = &rq->engine->breadcrumbs;
@@ -325,7 +327,8 @@ void i915_request_cancel_breadcrumb(struct i915_request *rq)
 	struct intel_breadcrumbs *b = &rq->engine->breadcrumbs;
 
 	lockdep_assert_held(&rq->lock);
-	lockdep_assert_irqs_disabled();
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT_FULL))
+		lockdep_assert_irqs_disabled();
 
 	/*
 	 * We must wait for b->irq_lock so that we know the interrupt handler
-- 
2.21.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

* Re: [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT Clark Williams
@ 2019-09-25 16:47   ` Sebastian Andrzej Siewior
  0 siblings, 0 replies; 10+ messages in thread
From: Sebastian Andrzej Siewior @ 2019-09-25 16:47 UTC (permalink / raw)
  To: Clark Williams; +Cc: bigeasy, tglx, linux-rt-users, linux-kernel

On 2019-08-19 19:33:17 [-0500], Clark Williams wrote:
> From: Clark Williams <williams@redhat.com>
> 
> The 'breadcrumb' code in the i915 driver calls lockdep_assert_irqs_disabled()
> when starting some operations. This is valid on a stock kernel
> but on a PREEMPT_RT kernel the spin_lock_irq*() calls to not disable
> interrupts and likewise the spin_unlock_irq*() calls to not enable interrupts.
> 
> Conditionalize these calls based on whether PREEMPT_RT_FULL is enabled.
> 
> Signed-off-by: Clark Williams <williams@redhat.com>
> ---
>  drivers/gpu/drm/i915/intel_breadcrumbs.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
> index 832cb6b1e9bd..3eef6010ebf6 100644
> --- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
> +++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
> @@ -101,7 +101,8 @@ __dma_fence_signal__notify(struct dma_fence *fence)
>  	struct dma_fence_cb *cur, *tmp;
>  
>  	lockdep_assert_held(fence->lock);
> -	lockdep_assert_irqs_disabled();
> +	if (!IS_ENABLED(CONFIG_PREEMPT_RT_FULL))
> +		lockdep_assert_irqs_disabled();
>  

Both asserts went in commit c36beba6b296b ("drm/i915: Seal races between
async GPU cancellation, retirement and signaling"). It looks like its
purpose is to ensure that ->lock is acquired and that interrupts are
also disabled because that lock is always acquired with disabled
interrupts. This isn't required, lockdep is clever enough to yell at you
if you manage to use it wrong.
Let me drop that lockdep_assert_irqs_disabled().

Sebastian

^ permalink raw reply	[flat|nested] 10+ messages in thread

* [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks
  2019-08-20  0:33 [PREEMPT_RT PATCH 0/3] i915 fixups for lockdep/lockdebugging Clark Williams
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT Clark Williams
@ 2019-08-20  0:33 ` Clark Williams
  2019-09-03  8:03   ` Sebastian Andrzej Siewior
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 3/3] i915: convert uncore lock to raw spinlock Clark Williams
  2 siblings, 1 reply; 10+ messages in thread
From: Clark Williams @ 2019-08-20  0:33 UTC (permalink / raw)
  To: bigeasy; +Cc: tglx, linux-rt-users, linux-kernel

From: Clark Williams <williams@redhat.com>

The following structures contain a member named 'irq_lock'.
These three locks are of type spinlock_t and are used in
multiple contexts including atomic:

    struct drm_i915_private
    struct intel_breadcrumbs
    strict intel_guc

Convert them all to be raw_spinlock_t so that lockdep and the lock
debugging code will be happy.

Signed-off-by: Clark Williams <williams@redhat.com>
---
 drivers/gpu/drm/i915/i915_debugfs.c        |   8 +-
 drivers/gpu/drm/i915/i915_drv.c            |   6 +-
 drivers/gpu/drm/i915/i915_drv.h            |   2 +-
 drivers/gpu/drm/i915/i915_irq.c            | 142 ++++++++++-----------
 drivers/gpu/drm/i915/intel_breadcrumbs.c   |  42 +++---
 drivers/gpu/drm/i915/intel_display.c       |   4 +-
 drivers/gpu/drm/i915/intel_engine_types.h  |   2 +-
 drivers/gpu/drm/i915/intel_fifo_underrun.c |  16 +--
 drivers/gpu/drm/i915/intel_guc.c           |   6 +-
 drivers/gpu/drm/i915/intel_guc.h           |  10 +-
 drivers/gpu/drm/i915/intel_hotplug.c       |  36 +++---
 drivers/gpu/drm/i915/intel_runtime_pm.c    |   8 +-
 drivers/gpu/drm/i915/intel_tv.c            |   8 +-
 13 files changed, 145 insertions(+), 145 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 5823ffb17821..969dd63b12e9 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -4336,12 +4336,12 @@ static ssize_t i915_hpd_storm_ctl_write(struct file *file,
 	else
 		DRM_DEBUG_KMS("Disabling HPD storm detection\n");
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	hotplug->hpd_storm_threshold = new_threshold;
 	/* Reset the HPD storm stats so we don't accidentally trigger a storm */
 	for_each_hpd_pin(i)
 		hotplug->stats[i].count = 0;
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* Re-enable hpd immediately if we were in an irq storm */
 	flush_delayed_work(&dev_priv->hotplug.reenable_work);
@@ -4414,12 +4414,12 @@ static ssize_t i915_hpd_short_storm_ctl_write(struct file *file,
 	DRM_DEBUG_KMS("%sabling HPD short storm detection\n",
 		      new_state ? "En" : "Dis");
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	hotplug->hpd_short_storm_enabled = new_state;
 	/* Reset the HPD storm stats so we don't accidentally trigger a storm */
 	for_each_hpd_pin(i)
 		hotplug->stats[i].count = 0;
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* Re-enable hpd immediately if we were in an irq storm */
 	flush_delayed_work(&dev_priv->hotplug.reenable_work);
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 1ad88e6d7c04..ce8d0d8d31de 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -879,7 +879,7 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv)
 
 	intel_uncore_init_early(&dev_priv->uncore);
 
-	spin_lock_init(&dev_priv->irq_lock);
+	raw_spin_lock_init(&dev_priv->irq_lock);
 	spin_lock_init(&dev_priv->gpu_error.lock);
 	mutex_init(&dev_priv->backlight_lock);
 
@@ -2219,10 +2219,10 @@ static int i915_drm_resume(struct drm_device *dev)
 	intel_modeset_init_hw(dev);
 	intel_init_clock_gating(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (dev_priv->display.hpd_irq_setup)
 		dev_priv->display.hpd_irq_setup(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	intel_dp_mst_resume(dev_priv);
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 066fd2a12851..293b69a870b9 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1552,7 +1552,7 @@ struct drm_i915_private {
 	struct resource mch_res;
 
 	/* protects the irq masks */
-	spinlock_t irq_lock;
+	raw_spinlock_t irq_lock;
 
 	bool display_irqs_enabled;
 
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 3d8abe9ebbb5..9b86926299b1 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -287,9 +287,9 @@ void i915_hotplug_interrupt_update(struct drm_i915_private *dev_priv,
 				   u32 mask,
 				   u32 bits)
 {
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	i915_hotplug_interrupt_update_locked(dev_priv, mask, bits);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 static u32
@@ -514,22 +514,22 @@ static void gen6_disable_pm_irq(struct drm_i915_private *dev_priv, u32 disable_m
 
 void gen11_reset_rps_interrupts(struct drm_i915_private *dev_priv)
 {
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	while (gen11_reset_one_iir(dev_priv, 0, GEN11_GTPM))
 		;
 
 	dev_priv->gt_pm.rps.pm_iir = 0;
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen6_reset_rps_interrupts(struct drm_i915_private *dev_priv)
 {
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	gen6_reset_pm_iir(dev_priv, GEN6_PM_RPS_EVENTS);
 	dev_priv->gt_pm.rps.pm_iir = 0;
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen6_enable_rps_interrupts(struct drm_i915_private *dev_priv)
@@ -539,7 +539,7 @@ void gen6_enable_rps_interrupts(struct drm_i915_private *dev_priv)
 	if (READ_ONCE(rps->interrupts_enabled))
 		return;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	WARN_ON_ONCE(rps->pm_iir);
 
 	if (INTEL_GEN(dev_priv) >= 11)
@@ -550,7 +550,7 @@ void gen6_enable_rps_interrupts(struct drm_i915_private *dev_priv)
 	rps->interrupts_enabled = true;
 	gen6_enable_pm_irq(dev_priv, dev_priv->pm_rps_events);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen6_disable_rps_interrupts(struct drm_i915_private *dev_priv)
@@ -560,14 +560,14 @@ void gen6_disable_rps_interrupts(struct drm_i915_private *dev_priv)
 	if (!READ_ONCE(rps->interrupts_enabled))
 		return;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	rps->interrupts_enabled = false;
 
 	I915_WRITE(GEN6_PMINTRMSK, gen6_sanitize_rps_pm_mask(dev_priv, ~0u));
 
 	gen6_disable_pm_irq(dev_priv, GEN6_PM_RPS_EVENTS);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 	synchronize_irq(dev_priv->drm.irq);
 
 	/* Now that we will not be generating any more work, flush any
@@ -586,35 +586,35 @@ void gen9_reset_guc_interrupts(struct drm_i915_private *dev_priv)
 {
 	assert_rpm_wakelock_held(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	gen6_reset_pm_iir(dev_priv, dev_priv->pm_guc_events);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen9_enable_guc_interrupts(struct drm_i915_private *dev_priv)
 {
 	assert_rpm_wakelock_held(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (!dev_priv->guc.interrupts_enabled) {
 		WARN_ON_ONCE(I915_READ(gen6_pm_iir(dev_priv)) &
 				       dev_priv->pm_guc_events);
 		dev_priv->guc.interrupts_enabled = true;
 		gen6_enable_pm_irq(dev_priv, dev_priv->pm_guc_events);
 	}
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen9_disable_guc_interrupts(struct drm_i915_private *dev_priv)
 {
 	assert_rpm_wakelock_held(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	dev_priv->guc.interrupts_enabled = false;
 
 	gen6_disable_pm_irq(dev_priv, dev_priv->pm_guc_events);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 	synchronize_irq(dev_priv->drm.irq);
 
 	gen9_reset_guc_interrupts(dev_priv);
@@ -813,14 +813,14 @@ static void i915_enable_asle_pipestat(struct drm_i915_private *dev_priv)
 	if (!i915_has_asle(dev_priv))
 		return;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	i915_enable_pipestat(dev_priv, PIPE_B, PIPE_LEGACY_BLC_EVENT_STATUS);
 	if (INTEL_GEN(dev_priv) >= 4)
 		i915_enable_pipestat(dev_priv, PIPE_A,
 				     PIPE_LEGACY_BLC_EVENT_STATUS);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 /*
@@ -1291,12 +1291,12 @@ static void gen6_pm_rps_work(struct work_struct *work)
 	int new_delay, adj, min, max;
 	u32 pm_iir = 0;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (rps->interrupts_enabled) {
 		pm_iir = fetch_and_zero(&rps->pm_iir);
 		client_boost = atomic_read(&rps->num_waiters);
 	}
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* Make sure we didn't queue anything we're not going to process. */
 	WARN_ON(pm_iir & ~dev_priv->pm_rps_events);
@@ -1373,10 +1373,10 @@ static void gen6_pm_rps_work(struct work_struct *work)
 
 out:
 	/* Make sure not to corrupt PMIMR state used by ringbuffer on GEN6 */
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (rps->interrupts_enabled)
 		gen6_unmask_pm_irq(dev_priv, dev_priv->pm_rps_events);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 
@@ -1454,9 +1454,9 @@ static void ivybridge_parity_work(struct work_struct *work)
 
 out:
 	WARN_ON(dev_priv->l3_parity.which_slice);
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	gen5_enable_gt_irq(dev_priv, GT_PARITY_ERROR(dev_priv));
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 }
@@ -1467,9 +1467,9 @@ static void ivybridge_parity_error_irq_handler(struct drm_i915_private *dev_priv
 	if (!HAS_L3_DPF(dev_priv))
 		return;
 
-	spin_lock(&dev_priv->irq_lock);
+	raw_spin_lock(&dev_priv->irq_lock);
 	gen5_disable_gt_irq(dev_priv, GT_PARITY_ERROR(dev_priv));
-	spin_unlock(&dev_priv->irq_lock);
+	raw_spin_unlock(&dev_priv->irq_lock);
 
 	iir &= GT_PARITY_ERROR(dev_priv);
 	if (iir & GT_RENDER_L3_PARITY_ERROR_INTERRUPT_S1)
@@ -1866,13 +1866,13 @@ static void gen6_rps_irq_handler(struct drm_i915_private *dev_priv, u32 pm_iir)
 	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 
 	if (pm_iir & dev_priv->pm_rps_events) {
-		spin_lock(&dev_priv->irq_lock);
+		raw_spin_lock(&dev_priv->irq_lock);
 		gen6_mask_pm_irq(dev_priv, pm_iir & dev_priv->pm_rps_events);
 		if (rps->interrupts_enabled) {
 			rps->pm_iir |= pm_iir & dev_priv->pm_rps_events;
 			schedule_work(&rps->work);
 		}
-		spin_unlock(&dev_priv->irq_lock);
+		raw_spin_unlock(&dev_priv->irq_lock);
 	}
 
 	if (INTEL_GEN(dev_priv) >= 8)
@@ -1909,10 +1909,10 @@ static void i9xx_pipestat_irq_ack(struct drm_i915_private *dev_priv,
 {
 	int pipe;
 
-	spin_lock(&dev_priv->irq_lock);
+	raw_spin_lock(&dev_priv->irq_lock);
 
 	if (!dev_priv->display_irqs_enabled) {
-		spin_unlock(&dev_priv->irq_lock);
+		raw_spin_unlock(&dev_priv->irq_lock);
 		return;
 	}
 
@@ -1966,7 +1966,7 @@ static void i9xx_pipestat_irq_ack(struct drm_i915_private *dev_priv,
 			I915_WRITE(reg, enable_mask);
 		}
 	}
-	spin_unlock(&dev_priv->irq_lock);
+	raw_spin_unlock(&dev_priv->irq_lock);
 }
 
 static void i8xx_pipestat_irq_handler(struct drm_i915_private *dev_priv,
@@ -3087,14 +3087,14 @@ gen11_gt_irq_handler(struct drm_i915_private * const i915,
 {
 	unsigned int bank;
 
-	spin_lock(&i915->irq_lock);
+	raw_spin_lock(&i915->irq_lock);
 
 	for (bank = 0; bank < 2; bank++) {
 		if (master_ctl & GEN11_GT_DW_IRQ(bank))
 			gen11_gt_bank_handler(i915, bank);
 	}
 
-	spin_unlock(&i915->irq_lock);
+	raw_spin_unlock(&i915->irq_lock);
 }
 
 static u32
@@ -3187,9 +3187,9 @@ static int i8xx_enable_vblank(struct drm_device *dev, unsigned int pipe)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	i915_enable_pipestat(dev_priv, pipe, PIPE_VBLANK_INTERRUPT_STATUS);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 
 	return 0;
 }
@@ -3209,10 +3209,10 @@ static int i965_enable_vblank(struct drm_device *dev, unsigned int pipe)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	i915_enable_pipestat(dev_priv, pipe,
 			     PIPE_START_VBLANK_INTERRUPT_STATUS);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 
 	return 0;
 }
@@ -3224,9 +3224,9 @@ static int ironlake_enable_vblank(struct drm_device *dev, unsigned int pipe)
 	u32 bit = INTEL_GEN(dev_priv) >= 7 ?
 		DE_PIPE_VBLANK_IVB(pipe) : DE_PIPE_VBLANK(pipe);
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	ilk_enable_display_irq(dev_priv, bit);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 
 	/* Even though there is no DMC, frame counter can get stuck when
 	 * PSR is active as no frames are generated.
@@ -3242,9 +3242,9 @@ static int gen8_enable_vblank(struct drm_device *dev, unsigned int pipe)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	bdw_enable_pipe_irq(dev_priv, pipe, GEN8_PIPE_VBLANK);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 
 	/* Even if there is no DMC, frame counter can get stuck when
 	 * PSR is active as no frames are generated, so check only for PSR.
@@ -3263,9 +3263,9 @@ static void i8xx_disable_vblank(struct drm_device *dev, unsigned int pipe)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	i915_disable_pipestat(dev_priv, pipe, PIPE_VBLANK_INTERRUPT_STATUS);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 }
 
 static void i945gm_disable_vblank(struct drm_device *dev, unsigned int pipe)
@@ -3283,10 +3283,10 @@ static void i965_disable_vblank(struct drm_device *dev, unsigned int pipe)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	i915_disable_pipestat(dev_priv, pipe,
 			      PIPE_START_VBLANK_INTERRUPT_STATUS);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 }
 
 static void ironlake_disable_vblank(struct drm_device *dev, unsigned int pipe)
@@ -3296,9 +3296,9 @@ static void ironlake_disable_vblank(struct drm_device *dev, unsigned int pipe)
 	u32 bit = INTEL_GEN(dev_priv) >= 7 ?
 		DE_PIPE_VBLANK_IVB(pipe) : DE_PIPE_VBLANK(pipe);
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	ilk_disable_display_irq(dev_priv, bit);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 }
 
 static void gen8_disable_vblank(struct drm_device *dev, unsigned int pipe)
@@ -3306,9 +3306,9 @@ static void gen8_disable_vblank(struct drm_device *dev, unsigned int pipe)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 	bdw_disable_pipe_irq(dev_priv, pipe, GEN8_PIPE_VBLANK);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 }
 
 static void i945gm_vblank_work_func(struct work_struct *work)
@@ -3486,10 +3486,10 @@ static void valleyview_irq_reset(struct drm_device *dev)
 
 	gen5_gt_irq_reset(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (dev_priv->display_irqs_enabled)
 		vlv_display_irq_reset(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 static void gen8_gt_irq_reset(struct drm_i915_private *dev_priv)
@@ -3583,10 +3583,10 @@ void gen8_irq_power_well_post_enable(struct drm_i915_private *dev_priv,
 	u32 extra_ier = GEN8_PIPE_VBLANK | GEN8_PIPE_FIFO_UNDERRUN;
 	enum pipe pipe;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	if (!intel_irqs_enabled(dev_priv)) {
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 		return;
 	}
 
@@ -3595,7 +3595,7 @@ void gen8_irq_power_well_post_enable(struct drm_i915_private *dev_priv,
 				  dev_priv->de_irq_mask[pipe],
 				  ~dev_priv->de_irq_mask[pipe] | extra_ier);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen8_irq_power_well_pre_disable(struct drm_i915_private *dev_priv,
@@ -3604,17 +3604,17 @@ void gen8_irq_power_well_pre_disable(struct drm_i915_private *dev_priv,
 	struct intel_uncore *uncore = &dev_priv->uncore;
 	enum pipe pipe;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	if (!intel_irqs_enabled(dev_priv)) {
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 		return;
 	}
 
 	for_each_pipe_masked(dev_priv, pipe, pipe_mask)
 		GEN8_IRQ_RESET_NDX(uncore, DE_PIPE, pipe);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* make sure we're done processing display irqs */
 	synchronize_irq(dev_priv->drm.irq);
@@ -3632,10 +3632,10 @@ static void cherryview_irq_reset(struct drm_device *dev)
 
 	GEN3_IRQ_RESET(uncore, GEN8_PCU_);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (dev_priv->display_irqs_enabled)
 		vlv_display_irq_reset(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 static u32 intel_hpd_enabled_irqs(struct drm_i915_private *dev_priv,
@@ -3997,9 +3997,9 @@ static int ironlake_irq_postinstall(struct drm_device *dev)
 		 * spinlocking not required here for correctness since interrupt
 		 * setup is guaranteed to run in single-threaded context. But we
 		 * need it to make the assert_spin_locked happy. */
-		spin_lock_irq(&dev_priv->irq_lock);
+		raw_spin_lock_irq(&dev_priv->irq_lock);
 		ilk_enable_display_irq(dev_priv, DE_PCU_EVENT);
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 	}
 
 	return 0;
@@ -4040,10 +4040,10 @@ static int valleyview_irq_postinstall(struct drm_device *dev)
 
 	gen5_gt_irq_postinstall(dev);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (dev_priv->display_irqs_enabled)
 		vlv_display_irq_postinstall(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	I915_WRITE(VLV_MASTER_IER, MASTER_INTERRUPT_ENABLE);
 	POSTING_READ(VLV_MASTER_IER);
@@ -4243,10 +4243,10 @@ static int cherryview_irq_postinstall(struct drm_device *dev)
 
 	gen8_gt_irq_postinstall(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (dev_priv->display_irqs_enabled)
 		vlv_display_irq_postinstall(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	I915_WRITE(GEN8_MASTER_IRQ, GEN8_MASTER_IRQ_CONTROL);
 	POSTING_READ(GEN8_MASTER_IRQ);
@@ -4289,10 +4289,10 @@ static int i8xx_irq_postinstall(struct drm_device *dev)
 
 	/* Interrupt setup is already guaranteed to be single-threaded, this is
 	 * just to make the assert_spin_locked check happy. */
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	i915_enable_pipestat(dev_priv, PIPE_A, PIPE_CRC_DONE_INTERRUPT_STATUS);
 	i915_enable_pipestat(dev_priv, PIPE_B, PIPE_CRC_DONE_INTERRUPT_STATUS);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	return 0;
 }
@@ -4467,10 +4467,10 @@ static int i915_irq_postinstall(struct drm_device *dev)
 
 	/* Interrupt setup is already guaranteed to be single-threaded, this is
 	 * just to make the assert_spin_locked check happy. */
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	i915_enable_pipestat(dev_priv, PIPE_A, PIPE_CRC_DONE_INTERRUPT_STATUS);
 	i915_enable_pipestat(dev_priv, PIPE_B, PIPE_CRC_DONE_INTERRUPT_STATUS);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	i915_enable_asle_pipestat(dev_priv);
 
@@ -4589,11 +4589,11 @@ static int i965_irq_postinstall(struct drm_device *dev)
 
 	/* Interrupt setup is already guaranteed to be single-threaded, this is
 	 * just to make the assert_spin_locked check happy. */
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	i915_enable_pipestat(dev_priv, PIPE_A, PIPE_GMBUS_INTERRUPT_STATUS);
 	i915_enable_pipestat(dev_priv, PIPE_A, PIPE_CRC_DONE_INTERRUPT_STATUS);
 	i915_enable_pipestat(dev_priv, PIPE_B, PIPE_CRC_DONE_INTERRUPT_STATUS);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	i915_enable_asle_pipestat(dev_priv);
 
diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
index 3eef6010ebf6..32ebbe887e61 100644
--- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
+++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
@@ -34,9 +34,9 @@ static void irq_enable(struct intel_engine_cs *engine)
 		return;
 
 	/* Caller disables interrupts */
-	spin_lock(&engine->i915->irq_lock);
+	raw_spin_lock(&engine->i915->irq_lock);
 	engine->irq_enable(engine);
-	spin_unlock(&engine->i915->irq_lock);
+	raw_spin_unlock(&engine->i915->irq_lock);
 }
 
 static void irq_disable(struct intel_engine_cs *engine)
@@ -45,9 +45,9 @@ static void irq_disable(struct intel_engine_cs *engine)
 		return;
 
 	/* Caller disables interrupts */
-	spin_lock(&engine->i915->irq_lock);
+	raw_spin_lock(&engine->i915->irq_lock);
 	engine->irq_disable(engine);
-	spin_unlock(&engine->i915->irq_lock);
+	raw_spin_unlock(&engine->i915->irq_lock);
 }
 
 static void __intel_breadcrumbs_disarm_irq(struct intel_breadcrumbs *b)
@@ -70,10 +70,10 @@ void intel_engine_disarm_breadcrumbs(struct intel_engine_cs *engine)
 	if (!b->irq_armed)
 		return;
 
-	spin_lock_irq(&b->irq_lock);
+	raw_spin_lock_irq(&b->irq_lock);
 	if (b->irq_armed)
 		__intel_breadcrumbs_disarm_irq(b);
-	spin_unlock_irq(&b->irq_lock);
+	raw_spin_unlock_irq(&b->irq_lock);
 }
 
 static inline bool __request_completed(const struct i915_request *rq)
@@ -119,7 +119,7 @@ void intel_engine_breadcrumbs_irq(struct intel_engine_cs *engine)
 	struct list_head *pos, *next;
 	LIST_HEAD(signal);
 
-	spin_lock(&b->irq_lock);
+	raw_spin_lock(&b->irq_lock);
 
 	if (b->irq_armed && list_empty(&b->signalers))
 		__intel_breadcrumbs_disarm_irq(b);
@@ -163,7 +163,7 @@ void intel_engine_breadcrumbs_irq(struct intel_engine_cs *engine)
 		}
 	}
 
-	spin_unlock(&b->irq_lock);
+	raw_spin_unlock(&b->irq_lock);
 
 	list_for_each_safe(pos, next, &signal) {
 		struct i915_request *rq =
@@ -198,22 +198,22 @@ void intel_engine_pin_breadcrumbs_irq(struct intel_engine_cs *engine)
 {
 	struct intel_breadcrumbs *b = &engine->breadcrumbs;
 
-	spin_lock_irq(&b->irq_lock);
+	raw_spin_lock_irq(&b->irq_lock);
 	if (!b->irq_enabled++)
 		irq_enable(engine);
 	GEM_BUG_ON(!b->irq_enabled); /* no overflow! */
-	spin_unlock_irq(&b->irq_lock);
+	raw_spin_unlock_irq(&b->irq_lock);
 }
 
 void intel_engine_unpin_breadcrumbs_irq(struct intel_engine_cs *engine)
 {
 	struct intel_breadcrumbs *b = &engine->breadcrumbs;
 
-	spin_lock_irq(&b->irq_lock);
+	raw_spin_lock_irq(&b->irq_lock);
 	GEM_BUG_ON(!b->irq_enabled); /* no underflow! */
 	if (!--b->irq_enabled)
 		irq_disable(engine);
-	spin_unlock_irq(&b->irq_lock);
+	raw_spin_unlock_irq(&b->irq_lock);
 }
 
 static void __intel_breadcrumbs_arm_irq(struct intel_breadcrumbs *b)
@@ -249,7 +249,7 @@ void intel_engine_init_breadcrumbs(struct intel_engine_cs *engine)
 {
 	struct intel_breadcrumbs *b = &engine->breadcrumbs;
 
-	spin_lock_init(&b->irq_lock);
+	raw_spin_lock_init(&b->irq_lock);
 	INIT_LIST_HEAD(&b->signalers);
 
 	init_irq_work(&b->irq_work, signal_irq_work);
@@ -260,14 +260,14 @@ void intel_engine_reset_breadcrumbs(struct intel_engine_cs *engine)
 	struct intel_breadcrumbs *b = &engine->breadcrumbs;
 	unsigned long flags;
 
-	spin_lock_irqsave(&b->irq_lock, flags);
+	raw_spin_lock_irqsave(&b->irq_lock, flags);
 
 	if (b->irq_enabled)
 		irq_enable(engine);
 	else
 		irq_disable(engine);
 
-	spin_unlock_irqrestore(&b->irq_lock, flags);
+	raw_spin_unlock_irqrestore(&b->irq_lock, flags);
 }
 
 void intel_engine_fini_breadcrumbs(struct intel_engine_cs *engine)
@@ -285,7 +285,7 @@ bool i915_request_enable_breadcrumb(struct i915_request *rq)
 		struct intel_context *ce = rq->hw_context;
 		struct list_head *pos;
 
-		spin_lock(&b->irq_lock);
+		raw_spin_lock(&b->irq_lock);
 		GEM_BUG_ON(test_bit(I915_FENCE_FLAG_SIGNAL, &rq->fence.flags));
 
 		__intel_breadcrumbs_arm_irq(b);
@@ -316,7 +316,7 @@ bool i915_request_enable_breadcrumb(struct i915_request *rq)
 			list_move_tail(&ce->signal_link, &b->signalers);
 
 		set_bit(I915_FENCE_FLAG_SIGNAL, &rq->fence.flags);
-		spin_unlock(&b->irq_lock);
+		raw_spin_unlock(&b->irq_lock);
 	}
 
 	return !__request_completed(rq);
@@ -336,7 +336,7 @@ void i915_request_cancel_breadcrumb(struct i915_request *rq)
 	 * the DMA_FENCE_FLAG_SIGNALED_BIT/I915_FENCE_FLAG_SIGNAL dance (if
 	 * required).
 	 */
-	spin_lock(&b->irq_lock);
+	raw_spin_lock(&b->irq_lock);
 	if (test_bit(I915_FENCE_FLAG_SIGNAL, &rq->fence.flags)) {
 		struct intel_context *ce = rq->hw_context;
 
@@ -346,7 +346,7 @@ void i915_request_cancel_breadcrumb(struct i915_request *rq)
 
 		clear_bit(I915_FENCE_FLAG_SIGNAL, &rq->fence.flags);
 	}
-	spin_unlock(&b->irq_lock);
+	raw_spin_unlock(&b->irq_lock);
 }
 
 void intel_engine_print_breadcrumbs(struct intel_engine_cs *engine,
@@ -361,7 +361,7 @@ void intel_engine_print_breadcrumbs(struct intel_engine_cs *engine,
 
 	drm_printf(p, "Signals:\n");
 
-	spin_lock_irq(&b->irq_lock);
+	raw_spin_lock_irq(&b->irq_lock);
 	list_for_each_entry(ce, &b->signalers, signal_link) {
 		list_for_each_entry(rq, &ce->signals, signal_link) {
 			drm_printf(p, "\t[%llx:%llx%s] @ %dms\n",
@@ -372,5 +372,5 @@ void intel_engine_print_breadcrumbs(struct intel_engine_cs *engine,
 				   jiffies_to_msecs(jiffies - rq->emitted_jiffies));
 		}
 	}
-	spin_unlock_irq(&b->irq_lock);
+	raw_spin_unlock_irq(&b->irq_lock);
 }
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 75105a2c59ea..6618a551fe09 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -3979,10 +3979,10 @@ void intel_finish_reset(struct drm_i915_private *dev_priv)
 		intel_modeset_init_hw(dev);
 		intel_init_clock_gating(dev_priv);
 
-		spin_lock_irq(&dev_priv->irq_lock);
+		raw_spin_lock_irq(&dev_priv->irq_lock);
 		if (dev_priv->display.hpd_irq_setup)
 			dev_priv->display.hpd_irq_setup(dev_priv);
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 		ret = __intel_display_resume(dev, state, ctx);
 		if (ret)
diff --git a/drivers/gpu/drm/i915/intel_engine_types.h b/drivers/gpu/drm/i915/intel_engine_types.h
index 4270ddb45f41..47e8e74b663e 100644
--- a/drivers/gpu/drm/i915/intel_engine_types.h
+++ b/drivers/gpu/drm/i915/intel_engine_types.h
@@ -307,7 +307,7 @@ struct intel_engine_cs {
 	 * the overhead of waking that client is much preferred.
 	 */
 	struct intel_breadcrumbs {
-		spinlock_t irq_lock;
+		raw_spinlock_t irq_lock;
 		struct list_head signalers;
 
 		struct irq_work irq_work; /* for use from inside irq_lock */
diff --git a/drivers/gpu/drm/i915/intel_fifo_underrun.c b/drivers/gpu/drm/i915/intel_fifo_underrun.c
index 74c8b0528294..9b6176f30e1e 100644
--- a/drivers/gpu/drm/i915/intel_fifo_underrun.c
+++ b/drivers/gpu/drm/i915/intel_fifo_underrun.c
@@ -293,10 +293,10 @@ bool intel_set_cpu_fifo_underrun_reporting(struct drm_i915_private *dev_priv,
 	unsigned long flags;
 	bool ret;
 
-	spin_lock_irqsave(&dev_priv->irq_lock, flags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, flags);
 	ret = __intel_set_cpu_fifo_underrun_reporting(&dev_priv->drm, pipe,
 						      enable);
-	spin_unlock_irqrestore(&dev_priv->irq_lock, flags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, flags);
 
 	return ret;
 }
@@ -333,7 +333,7 @@ bool intel_set_pch_fifo_underrun_reporting(struct drm_i915_private *dev_priv,
 	 * crtc on LPT won't cause issues.
 	 */
 
-	spin_lock_irqsave(&dev_priv->irq_lock, flags);
+	raw_spin_lock_irqsave(&dev_priv->irq_lock, flags);
 
 	old = !crtc->pch_fifo_underrun_disabled;
 	crtc->pch_fifo_underrun_disabled = !enable;
@@ -347,7 +347,7 @@ bool intel_set_pch_fifo_underrun_reporting(struct drm_i915_private *dev_priv,
 						pch_transcoder,
 						enable, old);
 
-	spin_unlock_irqrestore(&dev_priv->irq_lock, flags);
+	raw_spin_unlock_irqrestore(&dev_priv->irq_lock, flags);
 	return old;
 }
 
@@ -416,7 +416,7 @@ void intel_check_cpu_fifo_underruns(struct drm_i915_private *dev_priv)
 {
 	struct intel_crtc *crtc;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	for_each_intel_crtc(&dev_priv->drm, crtc) {
 		if (crtc->cpu_fifo_underrun_disabled)
@@ -428,7 +428,7 @@ void intel_check_cpu_fifo_underruns(struct drm_i915_private *dev_priv)
 			ivybridge_check_fifo_underruns(crtc);
 	}
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 /**
@@ -443,7 +443,7 @@ void intel_check_pch_fifo_underruns(struct drm_i915_private *dev_priv)
 {
 	struct intel_crtc *crtc;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	for_each_intel_crtc(&dev_priv->drm, crtc) {
 		if (crtc->pch_fifo_underrun_disabled)
@@ -453,5 +453,5 @@ void intel_check_pch_fifo_underruns(struct drm_i915_private *dev_priv)
 			cpt_check_pch_fifo_underruns(crtc);
 	}
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
diff --git a/drivers/gpu/drm/i915/intel_guc.c b/drivers/gpu/drm/i915/intel_guc.c
index 3aabfa2d9198..b15463b937b0 100644
--- a/drivers/gpu/drm/i915/intel_guc.c
+++ b/drivers/gpu/drm/i915/intel_guc.c
@@ -68,7 +68,7 @@ void intel_guc_init_early(struct intel_guc *guc)
 	intel_guc_log_init_early(&guc->log);
 
 	mutex_init(&guc->send_mutex);
-	spin_lock_init(&guc->irq_lock);
+	raw_spin_lock_init(&guc->irq_lock);
 	guc->send = intel_guc_send_nop;
 	guc->handler = intel_guc_to_host_event_handler_nop;
 	guc->notify = gen8_guc_raise_irq;
@@ -478,11 +478,11 @@ void intel_guc_to_host_event_handler_mmio(struct intel_guc *guc)
 	 * clears out the bit on handling the 1st interrupt.
 	 */
 	disable_rpm_wakeref_asserts(dev_priv);
-	spin_lock(&guc->irq_lock);
+	raw_spin_lock(&guc->irq_lock);
 	val = I915_READ(SOFT_SCRATCH(15));
 	msg = val & guc->msg_enabled_mask;
 	I915_WRITE(SOFT_SCRATCH(15), val & ~msg);
-	spin_unlock(&guc->irq_lock);
+	raw_spin_unlock(&guc->irq_lock);
 	enable_rpm_wakeref_asserts(dev_priv);
 
 	intel_guc_to_host_process_recv_msg(guc, &msg, 1);
diff --git a/drivers/gpu/drm/i915/intel_guc.h b/drivers/gpu/drm/i915/intel_guc.h
index 2c59ff8d9f39..cd05df72eca3 100644
--- a/drivers/gpu/drm/i915/intel_guc.h
+++ b/drivers/gpu/drm/i915/intel_guc.h
@@ -54,7 +54,7 @@ struct intel_guc {
 	struct drm_i915_gem_object *load_err_log;
 
 	/* intel_guc_recv interrupt related state */
-	spinlock_t irq_lock;
+	raw_spinlock_t irq_lock;
 	bool interrupts_enabled;
 	unsigned int msg_enabled_mask;
 
@@ -182,16 +182,16 @@ static inline int intel_guc_sanitize(struct intel_guc *guc)
 
 static inline void intel_guc_enable_msg(struct intel_guc *guc, u32 mask)
 {
-	spin_lock_irq(&guc->irq_lock);
+	raw_spin_lock_irq(&guc->irq_lock);
 	guc->msg_enabled_mask |= mask;
-	spin_unlock_irq(&guc->irq_lock);
+	raw_spin_unlock_irq(&guc->irq_lock);
 }
 
 static inline void intel_guc_disable_msg(struct intel_guc *guc, u32 mask)
 {
-	spin_lock_irq(&guc->irq_lock);
+	raw_spin_lock_irq(&guc->irq_lock);
 	guc->msg_enabled_mask &= ~mask;
-	spin_unlock_irq(&guc->irq_lock);
+	raw_spin_unlock_irq(&guc->irq_lock);
 }
 
 int intel_guc_reset_engine(struct intel_guc *guc,
diff --git a/drivers/gpu/drm/i915/intel_hotplug.c b/drivers/gpu/drm/i915/intel_hotplug.c
index b8937c788f03..188633bd8c08 100644
--- a/drivers/gpu/drm/i915/intel_hotplug.c
+++ b/drivers/gpu/drm/i915/intel_hotplug.c
@@ -231,7 +231,7 @@ static void intel_hpd_irq_storm_reenable_work(struct work_struct *work)
 
 	wakeref = intel_runtime_pm_get(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	for_each_hpd_pin(pin) {
 		struct drm_connector *connector;
 		struct drm_connector_list_iter conn_iter;
@@ -260,7 +260,7 @@ static void intel_hpd_irq_storm_reenable_work(struct work_struct *work)
 	}
 	if (dev_priv->display_irqs_enabled && dev_priv->display.hpd_irq_setup)
 		dev_priv->display.hpd_irq_setup(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	intel_runtime_pm_put(dev_priv, wakeref);
 }
@@ -303,12 +303,12 @@ static void i915_digport_work_func(struct work_struct *work)
 	struct intel_encoder *encoder;
 	u32 old_bits = 0;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	long_port_mask = dev_priv->hotplug.long_port_mask;
 	dev_priv->hotplug.long_port_mask = 0;
 	short_port_mask = dev_priv->hotplug.short_port_mask;
 	dev_priv->hotplug.short_port_mask = 0;
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	for_each_intel_encoder(&dev_priv->drm, encoder) {
 		struct intel_digital_port *dig_port;
@@ -335,9 +335,9 @@ static void i915_digport_work_func(struct work_struct *work)
 	}
 
 	if (old_bits) {
-		spin_lock_irq(&dev_priv->irq_lock);
+		raw_spin_lock_irq(&dev_priv->irq_lock);
 		dev_priv->hotplug.event_bits |= old_bits;
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 		schedule_work(&dev_priv->hotplug.hotplug_work);
 	}
 }
@@ -360,7 +360,7 @@ static void i915_hotplug_work_func(struct work_struct *work)
 	mutex_lock(&dev->mode_config.mutex);
 	DRM_DEBUG_KMS("running encoder hotplug functions\n");
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	hpd_event_bits = dev_priv->hotplug.event_bits;
 	dev_priv->hotplug.event_bits = 0;
@@ -368,7 +368,7 @@ static void i915_hotplug_work_func(struct work_struct *work)
 	/* Enable polling for connectors which had HPD IRQ storms */
 	intel_hpd_irq_storm_switch_to_polling(dev_priv);
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	drm_connector_list_iter_begin(dev, &conn_iter);
 	drm_for_each_connector_iter(connector, &conn_iter) {
@@ -421,7 +421,7 @@ void intel_hpd_irq_handler(struct drm_i915_private *dev_priv,
 	if (!pin_mask)
 		return;
 
-	spin_lock(&dev_priv->irq_lock);
+	raw_spin_lock(&dev_priv->irq_lock);
 
 	/*
 	 * Determine whether ->hpd_pulse() exists for each pin, and
@@ -504,7 +504,7 @@ void intel_hpd_irq_handler(struct drm_i915_private *dev_priv,
 	 */
 	if (storm_detected && dev_priv->display_irqs_enabled)
 		dev_priv->display.hpd_irq_setup(dev_priv);
-	spin_unlock(&dev_priv->irq_lock);
+	raw_spin_unlock(&dev_priv->irq_lock);
 
 	/*
 	 * Our hotplug handler can grab modeset locks (by calling down into the
@@ -549,10 +549,10 @@ void intel_hpd_init(struct drm_i915_private *dev_priv)
 	 * just to make the assert_spin_locked checks happy.
 	 */
 	if (dev_priv->display_irqs_enabled && dev_priv->display.hpd_irq_setup) {
-		spin_lock_irq(&dev_priv->irq_lock);
+		raw_spin_lock_irq(&dev_priv->irq_lock);
 		if (dev_priv->display_irqs_enabled)
 			dev_priv->display.hpd_irq_setup(dev_priv);
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 	}
 }
 
@@ -644,13 +644,13 @@ void intel_hpd_init_work(struct drm_i915_private *dev_priv)
 
 void intel_hpd_cancel_work(struct drm_i915_private *dev_priv)
 {
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 
 	dev_priv->hotplug.long_port_mask = 0;
 	dev_priv->hotplug.short_port_mask = 0;
 	dev_priv->hotplug.event_bits = 0;
 
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	cancel_work_sync(&dev_priv->hotplug.dig_port_work);
 	cancel_work_sync(&dev_priv->hotplug.hotplug_work);
@@ -665,12 +665,12 @@ bool intel_hpd_disable(struct drm_i915_private *dev_priv, enum hpd_pin pin)
 	if (pin == HPD_NONE)
 		return false;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	if (dev_priv->hotplug.stats[pin].state == HPD_ENABLED) {
 		dev_priv->hotplug.stats[pin].state = HPD_DISABLED;
 		ret = true;
 	}
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	return ret;
 }
@@ -680,7 +680,7 @@ void intel_hpd_enable(struct drm_i915_private *dev_priv, enum hpd_pin pin)
 	if (pin == HPD_NONE)
 		return;
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	dev_priv->hotplug.stats[pin].state = HPD_ENABLED;
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 }
diff --git a/drivers/gpu/drm/i915/intel_runtime_pm.c b/drivers/gpu/drm/i915/intel_runtime_pm.c
index 6150e35bf7b5..11c0f2a738cd 100644
--- a/drivers/gpu/drm/i915/intel_runtime_pm.c
+++ b/drivers/gpu/drm/i915/intel_runtime_pm.c
@@ -1324,9 +1324,9 @@ static void vlv_display_power_well_init(struct drm_i915_private *dev_priv)
 
 	vlv_init_display_clock_gating(dev_priv);
 
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	valleyview_enable_display_irqs(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	/*
 	 * During driver initialization/resume we can avoid restoring the
@@ -1350,9 +1350,9 @@ static void vlv_display_power_well_init(struct drm_i915_private *dev_priv)
 
 static void vlv_display_power_well_deinit(struct drm_i915_private *dev_priv)
 {
-	spin_lock_irq(&dev_priv->irq_lock);
+	raw_spin_lock_irq(&dev_priv->irq_lock);
 	valleyview_disable_display_irqs(dev_priv);
-	spin_unlock_irq(&dev_priv->irq_lock);
+	raw_spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* make sure we're done processing display irqs */
 	synchronize_irq(dev_priv->drm.irq);
diff --git a/drivers/gpu/drm/i915/intel_tv.c b/drivers/gpu/drm/i915/intel_tv.c
index 5dbba33f4202..bb7dd33935a1 100644
--- a/drivers/gpu/drm/i915/intel_tv.c
+++ b/drivers/gpu/drm/i915/intel_tv.c
@@ -1574,11 +1574,11 @@ intel_tv_detect_type(struct intel_tv *intel_tv,
 
 	/* Disable TV interrupts around load detect or we'll recurse */
 	if (connector->polled & DRM_CONNECTOR_POLL_HPD) {
-		spin_lock_irq(&dev_priv->irq_lock);
+		raw_spin_lock_irq(&dev_priv->irq_lock);
 		i915_disable_pipestat(dev_priv, 0,
 				      PIPE_HOTPLUG_INTERRUPT_STATUS |
 				      PIPE_HOTPLUG_TV_INTERRUPT_STATUS);
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 	}
 
 	save_tv_dac = tv_dac = I915_READ(TV_DAC);
@@ -1646,11 +1646,11 @@ intel_tv_detect_type(struct intel_tv *intel_tv,
 
 	/* Restore interrupt config */
 	if (connector->polled & DRM_CONNECTOR_POLL_HPD) {
-		spin_lock_irq(&dev_priv->irq_lock);
+		raw_spin_lock_irq(&dev_priv->irq_lock);
 		i915_enable_pipestat(dev_priv, 0,
 				     PIPE_HOTPLUG_INTERRUPT_STATUS |
 				     PIPE_HOTPLUG_TV_INTERRUPT_STATUS);
-		spin_unlock_irq(&dev_priv->irq_lock);
+		raw_spin_unlock_irq(&dev_priv->irq_lock);
 	}
 
 	return type;
-- 
2.21.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

* Re: [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks Clark Williams
@ 2019-09-03  8:03   ` Sebastian Andrzej Siewior
  2019-09-03 13:09     ` Sebastian Andrzej Siewior
  2019-09-16 21:21     ` Sean V Kelley
  0 siblings, 2 replies; 10+ messages in thread
From: Sebastian Andrzej Siewior @ 2019-09-03  8:03 UTC (permalink / raw)
  To: Clark Williams; +Cc: bigeasy, tglx, linux-rt-users, linux-kernel

On 2019-08-19 19:33:18 [-0500], Clark Williams wrote:
> From: Clark Williams <williams@redhat.com>
> 
> The following structures contain a member named 'irq_lock'.
> These three locks are of type spinlock_t and are used in
> multiple contexts including atomic:
> 
>     struct drm_i915_private
>     struct intel_breadcrumbs
>     strict intel_guc
> 
> Convert them all to be raw_spinlock_t so that lockdep and the lock
> debugging code will be happy.

What is your motivation to make the lock raw?
I did the following:

 void intel_engine_signal_breadcrumbs(struct intel_engine_cs *engine)
 {
-       local_irq_disable();
-       intel_engine_breadcrumbs_irq(engine);
-       local_irq_enable();
+       if (IS_ENABLED(CONFIG_PREEMPT_RT_FULL)) {
+               intel_engine_breadcrumbs_irq(engine);
+       } else {
+               local_irq_disable();
+               intel_engine_breadcrumbs_irq(engine);
+               local_irq_enable();
+       }
 }

and lockdep was quiet (+ ignoring/patching the lockdep-irq-off-asserts).
The local_irq_disable() is here (my interpretation of the situation)
because that function is called from process context while the remaining
callers invoke intel_engine_breadcrumbs_irq() from the interrupt
handler and it acquires irq_lock via a plain spin_lock().  That
local_irq_disable() would be required if everyone did a _irqsave().

I tried to check how much worse the latency gets here but I didn't see
anything in a brief test. What I saw however is that switching to
fullscreen while playing a video gives me ~0.5 to ~2ms latency. This is
has nothing to do with this change, I have to dig deeper… It might be
one of the preempt_disable() section I just noticed.
I would prefer to keep the lock non-raw unless there is actual need for
it.

Sebastian

^ permalink raw reply	[flat|nested] 10+ messages in thread

* Re: [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks
  2019-09-03  8:03   ` Sebastian Andrzej Siewior
@ 2019-09-03 13:09     ` Sebastian Andrzej Siewior
  2019-09-16 21:21     ` Sean V Kelley
  1 sibling, 0 replies; 10+ messages in thread
From: Sebastian Andrzej Siewior @ 2019-09-03 13:09 UTC (permalink / raw)
  To: Clark Williams; +Cc: bigeasy, tglx, linux-rt-users, linux-kernel

On 2019-09-03 10:03:35 [+0200], To Clark Williams wrote:
> anything in a brief test. What I saw however is that switching to
> fullscreen while playing a video gives me ~0.5 to ~2ms latency. This is
> has nothing to do with this change, I have to dig deeper… It might be
> one of the preempt_disable() section I just noticed.
> I would prefer to keep the lock non-raw unless there is actual need for
> it.

That latency spike is caused by the iommu, qi_submit_sync() to be exact.
Without the iomm I don't see much difference between the two patches…
 
Sebastian

^ permalink raw reply	[flat|nested] 10+ messages in thread

* Re: [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks
  2019-09-03  8:03   ` Sebastian Andrzej Siewior
  2019-09-03 13:09     ` Sebastian Andrzej Siewior
@ 2019-09-16 21:21     ` Sean V Kelley
  2019-09-27 12:29       ` Sebastian Andrzej Siewior
  1 sibling, 1 reply; 10+ messages in thread
From: Sean V Kelley @ 2019-09-16 21:21 UTC (permalink / raw)
  To: Sebastian Andrzej Siewior
  Cc: Clark Williams, bigeasy, tglx, linux-rt-users, linux-kernel



> On Sep 3, 2019, at 1:03 AM, Sebastian Andrzej Siewior <bigeasy@linutronix.de> wrote:
> 
> On 2019-08-19 19:33:18 [-0500], Clark Williams wrote:
>> From: Clark Williams <williams@redhat.com>
>> 
>> The following structures contain a member named 'irq_lock'.
>> These three locks are of type spinlock_t and are used in
>> multiple contexts including atomic:
>> 
>>    struct drm_i915_private
>>    struct intel_breadcrumbs
>>    strict intel_guc
>> 
>> Convert them all to be raw_spinlock_t so that lockdep and the lock
>> debugging code will be happy.
> 
> What is your motivation to make the lock raw?
> I did the following:
> 
> void intel_engine_signal_breadcrumbs(struct intel_engine_cs *engine)
> {
> -       local_irq_disable();
> -       intel_engine_breadcrumbs_irq(engine);
> -       local_irq_enable();
> +       if (IS_ENABLED(CONFIG_PREEMPT_RT_FULL)) {
> +               intel_engine_breadcrumbs_irq(engine);
> +       } else {
> +               local_irq_disable();
> +               intel_engine_breadcrumbs_irq(engine);
> +               local_irq_enable();
> +       }
> }
> 
> and lockdep was quiet (+ ignoring/patching the lockdep-irq-off-asserts).
> The local_irq_disable() is here (my interpretation of the situation)
> because that function is called from process context while the remaining
> callers invoke intel_engine_breadcrumbs_irq() from the interrupt
> handler and it acquires irq_lock via a plain spin_lock().  That
> local_irq_disable() would be required if everyone did a _irqsave().

I’ve tested this also on the v5.2.14-rt7 and can confirm that it avoids the need for making the locks raw.

Tested-by: Sean V Kelley <sean.v.kelley@linux.intel.com>

Thanks,

Sean

> 
> I tried to check how much worse the latency gets here but I didn't see
> anything in a brief test. What I saw however is that switching to
> fullscreen while playing a video gives me ~0.5 to ~2ms latency. This is
> has nothing to do with this change, I have to dig deeper… It might be
> one of the preempt_disable() section I just noticed.
> I would prefer to keep the lock non-raw unless there is actual need for
> it.
> 
> Sebastian


^ permalink raw reply	[flat|nested] 10+ messages in thread

* Re: [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks
  2019-09-16 21:21     ` Sean V Kelley
@ 2019-09-27 12:29       ` Sebastian Andrzej Siewior
       [not found]         ` <CAPAFJkofsYgSOxzM4JDkOV3Ra0C-AEpU2eZJjR9F9e_qv-6=rg@mail.gmail.com>
  0 siblings, 1 reply; 10+ messages in thread
From: Sebastian Andrzej Siewior @ 2019-09-27 12:29 UTC (permalink / raw)
  To: Sean V Kelley; +Cc: Clark Williams, tglx, linux-rt-users, linux-kernel

On 2019-09-16 14:21:22 [-0700], Sean V Kelley wrote:
> I’ve tested this also on the v5.2.14-rt7 and can confirm that it avoids the need for making the locks raw.
> 
> Tested-by: Sean V Kelley <sean.v.kelley@linux.intel.com>

Good. I went for an alternative approach in v5.2.17-rt9. Now I believe
that the three here are obsolete. If not, please give a sign.

> Thanks,
> 
> Sean

Sebastian

^ permalink raw reply	[flat|nested] 10+ messages in thread

[parent not found: <CAPAFJkofsYgSOxzM4JDkOV3Ra0C-AEpU2eZJjR9F9e_qv-6=rg@mail.gmail.com>]

* Re: [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks
       [not found]         ` <CAPAFJkofsYgSOxzM4JDkOV3Ra0C-AEpU2eZJjR9F9e_qv-6=rg@mail.gmail.com>
@ 2019-09-27 15:18           ` Sean V Kelley
  0 siblings, 0 replies; 10+ messages in thread
From: Sean V Kelley @ 2019-09-27 15:18 UTC (permalink / raw)
  To: Clark Williams
  Cc: Sebastian Andrzej Siewior, Thomas Gleixner, RT, linux-kernel



> On Sep 27, 2019, at 7:58 AM, Clark Williams <clark.williams@gmail.com> wrote:
> 
> I'm running v5.2.17-rt9 now and have not seen any i915 related splats. I think we're good to drop my patches
> 
> On Fri, Sep 27, 2019 at 7:29 AM Sebastian Andrzej Siewior <bigeasy@linutronix.de> wrote:
> On 2019-09-16 14:21:22 [-0700], Sean V Kelley wrote:
> > I’ve tested this also on the v5.2.14-rt7 and can confirm that it avoids the need for making the locks raw.
> > 
> > Tested-by: Sean V Kelley <sean.v.kelley@linux.intel.com>
> 
> Good. I went for an alternative approach in v5.2.17-rt9. Now I believe
> that the three here are obsolete. If not, please give a sign.

Confirmed as well..

Sean

> 
> > Thanks,
> > 
> > Sean
> 
> Sebastian


^ permalink raw reply	[flat|nested] 10+ messages in thread

* [PREEMPT_RT PATCH 3/3] i915: convert uncore lock to raw spinlock
  2019-08-20  0:33 [PREEMPT_RT PATCH 0/3] i915 fixups for lockdep/lockdebugging Clark Williams
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT Clark Williams
  2019-08-20  0:33 ` [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks Clark Williams
@ 2019-08-20  0:33 ` Clark Williams
  2 siblings, 0 replies; 10+ messages in thread
From: Clark Williams @ 2019-08-20  0:33 UTC (permalink / raw)
  To: bigeasy; +Cc: tglx, linux-rt-users, linux-kernel

From: Clark Williams <williams@redhat.com>

The structure intel_uncore contains a spinlock member
named 'lock' which is used in multiple contexts. Convert
it to a raw spinlock so that lockdep and the lock debugging
code will be happy.

Signed-off-by: Clark Williams <williams@redhat.com>
---
 drivers/gpu/drm/i915/i915_gem.c          |  4 +-
 drivers/gpu/drm/i915/i915_irq.c          | 12 +++---
 drivers/gpu/drm/i915/i915_pmu.c          |  4 +-
 drivers/gpu/drm/i915/intel_display.c     | 16 ++++----
 drivers/gpu/drm/i915/intel_engine_cs.c   |  4 +-
 drivers/gpu/drm/i915/intel_pm.c          |  8 ++--
 drivers/gpu/drm/i915/intel_sprite.c      | 32 +++++++--------
 drivers/gpu/drm/i915/intel_uncore.c      | 52 ++++++++++++------------
 drivers/gpu/drm/i915/intel_uncore.h      |  2 +-
 drivers/gpu/drm/i915/intel_workarounds.c |  4 +-
 10 files changed, 69 insertions(+), 69 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 2910a133077a..09e097f8aae2 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -726,11 +726,11 @@ void i915_gem_flush_ggtt_writes(struct drm_i915_private *dev_priv)
 	i915_gem_chipset_flush(dev_priv);
 
 	with_intel_runtime_pm(dev_priv, wakeref) {
-		spin_lock_irq(&dev_priv->uncore.lock);
+		raw_spin_lock_irq(&dev_priv->uncore.lock);
 
 		POSTING_READ_FW(RING_HEAD(RENDER_RING_BASE));
 
-		spin_unlock_irq(&dev_priv->uncore.lock);
+		raw_spin_unlock_irq(&dev_priv->uncore.lock);
 	}
 }
 
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 9b86926299b1..b19a630e78c7 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -914,7 +914,7 @@ static u32 i915_get_vblank_counter(struct drm_device *dev, unsigned int pipe)
 	high_frame = PIPEFRAME(pipe);
 	low_frame = PIPEFRAMEPIXEL(pipe);
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	/*
 	 * High & low register fields aren't synchronized, so make sure
@@ -927,7 +927,7 @@ static u32 i915_get_vblank_counter(struct drm_device *dev, unsigned int pipe)
 		high2 = I915_READ_FW(high_frame) & PIPE_FRAME_HIGH_MASK;
 	} while (high1 != high2);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 
 	high1 >>= PIPE_FRAME_HIGH_SHIFT;
 	pixel = low & PIPE_PIXEL_MASK;
@@ -1097,7 +1097,7 @@ static bool i915_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 	 * register reads, potentially with preemption disabled, so the
 	 * following code must not block on uncore.lock.
 	 */
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	/* preempt_disable_rt() should go right here in PREEMPT_RT patchset. */
 	preempt_disable_rt();
@@ -1154,7 +1154,7 @@ static bool i915_get_crtc_scanoutpos(struct drm_device *dev, unsigned int pipe,
 	/* preempt_enable_rt() should go right here in PREEMPT_RT patchset. */
 	preempt_enable_rt();
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 
 	/*
 	 * While in vblank, position will be negative
@@ -1184,9 +1184,9 @@ int intel_get_crtc_scanline(struct intel_crtc *crtc)
 	unsigned long irqflags;
 	int position;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 	position = __intel_get_crtc_scanline(crtc);
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 
 	return position;
 }
diff --git a/drivers/gpu/drm/i915/i915_pmu.c b/drivers/gpu/drm/i915/i915_pmu.c
index 46a52da3db29..1df197c160dc 100644
--- a/drivers/gpu/drm/i915/i915_pmu.c
+++ b/drivers/gpu/drm/i915/i915_pmu.c
@@ -172,7 +172,7 @@ engines_sample(struct drm_i915_private *dev_priv, unsigned int period_ns)
 	if (!wakeref)
 		return;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, flags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, flags);
 	for_each_engine(engine, dev_priv, id) {
 		struct intel_engine_pmu *pmu = &engine->pmu;
 		bool busy;
@@ -202,7 +202,7 @@ engines_sample(struct drm_i915_private *dev_priv, unsigned int period_ns)
 		if (busy)
 			add_sample(&pmu->sample[I915_SAMPLE_BUSY], period_ns);
 	}
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, flags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, flags);
 
 	intel_runtime_pm_put(dev_priv, wakeref);
 }
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 6618a551fe09..1994876db6a2 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -3420,7 +3420,7 @@ static void i9xx_update_plane(struct intel_plane *plane,
 	else
 		dspaddr_offset = linear_offset;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(DSPSTRIDE(i9xx_plane), plane_state->color_plane[0].stride);
 
@@ -3462,7 +3462,7 @@ static void i9xx_update_plane(struct intel_plane *plane,
 			      intel_plane_ggtt_offset(plane_state) +
 			      dspaddr_offset);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void i9xx_disable_plane(struct intel_plane *plane,
@@ -3485,7 +3485,7 @@ static void i9xx_disable_plane(struct intel_plane *plane,
 	 */
 	dspcntr = i9xx_plane_ctl_crtc(crtc_state);
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(DSPCNTR(i9xx_plane), dspcntr);
 	if (INTEL_GEN(dev_priv) >= 4)
@@ -3493,7 +3493,7 @@ static void i9xx_disable_plane(struct intel_plane *plane,
 	else
 		I915_WRITE_FW(DSPADDR(i9xx_plane), 0);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static bool i9xx_plane_get_hw_state(struct intel_plane *plane,
@@ -10282,7 +10282,7 @@ static void i845_update_cursor(struct intel_plane *plane,
 		pos = intel_cursor_position(plane_state);
 	}
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	/* On these chipsets we can only modify the base/size/stride
 	 * whilst the cursor is disabled.
@@ -10303,7 +10303,7 @@ static void i845_update_cursor(struct intel_plane *plane,
 		I915_WRITE_FW(CURPOS(PIPE_A), pos);
 	}
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void i845_disable_cursor(struct intel_plane *plane,
@@ -10507,7 +10507,7 @@ static void i9xx_update_cursor(struct intel_plane *plane,
 		pos = intel_cursor_position(plane_state);
 	}
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	/*
 	 * On some platforms writing CURCNTR first will also
@@ -10549,7 +10549,7 @@ static void i9xx_update_cursor(struct intel_plane *plane,
 		I915_WRITE_FW(CURBASE(pipe), base);
 	}
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void i9xx_disable_cursor(struct intel_plane *plane,
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index 9d4f12e982c3..2e41b645d459 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -932,7 +932,7 @@ read_subslice_reg(struct drm_i915_private *dev_priv, int slice,
 						     GEN8_MCR_SELECTOR,
 						     FW_REG_READ | FW_REG_WRITE);
 
-	spin_lock_irq(&uncore->lock);
+	raw_spin_lock_irq(&uncore->lock);
 	intel_uncore_forcewake_get__locked(uncore, fw_domains);
 
 	mcr = intel_uncore_read_fw(uncore, GEN8_MCR_SELECTOR);
@@ -952,7 +952,7 @@ read_subslice_reg(struct drm_i915_private *dev_priv, int slice,
 	intel_uncore_write_fw(uncore, GEN8_MCR_SELECTOR, mcr);
 
 	intel_uncore_forcewake_put__locked(uncore, fw_domains);
-	spin_unlock_irq(&uncore->lock);
+	raw_spin_unlock_irq(&uncore->lock);
 
 	return ret;
 }
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 44be676fabd6..3911aa3e825b 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -1971,7 +1971,7 @@ static void vlv_atomic_update_fifo(struct intel_atomic_state *state,
 	 * intel_pipe_update_start() has already disabled interrupts
 	 * for us, so a plain spin_lock() is sufficient here.
 	 */
-	spin_lock(&dev_priv->uncore.lock);
+	raw_spin_lock(&dev_priv->uncore.lock);
 
 	switch (crtc->pipe) {
 		u32 dsparb, dsparb2, dsparb3;
@@ -2032,7 +2032,7 @@ static void vlv_atomic_update_fifo(struct intel_atomic_state *state,
 
 	POSTING_READ_FW(DSPARB);
 
-	spin_unlock(&dev_priv->uncore.lock);
+	raw_spin_unlock(&dev_priv->uncore.lock);
 }
 
 #undef VLV_FIFO
@@ -10060,7 +10060,7 @@ u64 intel_rc6_residency_ns(struct drm_i915_private *dev_priv,
 
 	fw_domains = intel_uncore_forcewake_for_reg(uncore, reg, FW_REG_READ);
 
-	spin_lock_irqsave(&uncore->lock, flags);
+	raw_spin_lock_irqsave(&uncore->lock, flags);
 	intel_uncore_forcewake_get__locked(uncore, fw_domains);
 
 	/* On VLV and CHV, residency time is in CZ units rather than 1.28us */
@@ -10103,7 +10103,7 @@ u64 intel_rc6_residency_ns(struct drm_i915_private *dev_priv,
 	dev_priv->gt_pm.rc6.cur_residency[i] = time_hw;
 
 	intel_uncore_forcewake_put__locked(uncore, fw_domains);
-	spin_unlock_irqrestore(&uncore->lock, flags);
+	raw_spin_unlock_irqrestore(&uncore->lock, flags);
 
 	return mul_u64_u32_div(time_hw, mul, div);
 }
diff --git a/drivers/gpu/drm/i915/intel_sprite.c b/drivers/gpu/drm/i915/intel_sprite.c
index eb4b0b31f6f0..87427c3b588a 100644
--- a/drivers/gpu/drm/i915/intel_sprite.c
+++ b/drivers/gpu/drm/i915/intel_sprite.c
@@ -541,7 +541,7 @@ skl_program_plane(struct intel_plane *plane,
 		crtc_y = 0;
 	}
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(PLANE_STRIDE(pipe, plane_id), stride);
 	I915_WRITE_FW(PLANE_POS(pipe, plane_id), (crtc_y << 16) | crtc_x);
@@ -601,7 +601,7 @@ skl_program_plane(struct intel_plane *plane,
 	if (!slave && plane_state->scaler_id >= 0)
 		skl_program_scaler(plane, crtc_state, plane_state);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void
@@ -638,7 +638,7 @@ skl_disable_plane(struct intel_plane *plane,
 	enum pipe pipe = plane->pipe;
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	if (icl_is_hdr_plane(dev_priv, plane_id))
 		I915_WRITE_FW(PLANE_CUS_CTL(pipe, plane_id), 0);
@@ -648,7 +648,7 @@ skl_disable_plane(struct intel_plane *plane,
 	I915_WRITE_FW(PLANE_CTL(pipe, plane_id), 0);
 	I915_WRITE_FW(PLANE_SURF(pipe, plane_id), 0);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static bool
@@ -878,7 +878,7 @@ vlv_update_plane(struct intel_plane *plane,
 
 	linear_offset = intel_fb_xy_to_linear(x, y, plane_state, 0);
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(SPSTRIDE(pipe, plane_id),
 		      plane_state->color_plane[0].stride);
@@ -909,7 +909,7 @@ vlv_update_plane(struct intel_plane *plane,
 
 	vlv_update_clrc(plane_state);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void
@@ -921,12 +921,12 @@ vlv_disable_plane(struct intel_plane *plane,
 	enum plane_id plane_id = plane->id;
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(SPCNTR(pipe, plane_id), 0);
 	I915_WRITE_FW(SPSURF(pipe, plane_id), 0);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static bool
@@ -1059,7 +1059,7 @@ ivb_update_plane(struct intel_plane *plane,
 
 	linear_offset = intel_fb_xy_to_linear(x, y, plane_state, 0);
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(SPRSTRIDE(pipe), plane_state->color_plane[0].stride);
 	I915_WRITE_FW(SPRPOS(pipe), (crtc_y << 16) | crtc_x);
@@ -1091,7 +1091,7 @@ ivb_update_plane(struct intel_plane *plane,
 	I915_WRITE_FW(SPRSURF(pipe),
 		      intel_plane_ggtt_offset(plane_state) + sprsurf_offset);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void
@@ -1102,7 +1102,7 @@ ivb_disable_plane(struct intel_plane *plane,
 	enum pipe pipe = plane->pipe;
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(SPRCTL(pipe), 0);
 	/* Disable the scaler */
@@ -1110,7 +1110,7 @@ ivb_disable_plane(struct intel_plane *plane,
 		I915_WRITE_FW(SPRSCALE(pipe), 0);
 	I915_WRITE_FW(SPRSURF(pipe), 0);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static bool
@@ -1250,7 +1250,7 @@ g4x_update_plane(struct intel_plane *plane,
 
 	linear_offset = intel_fb_xy_to_linear(x, y, plane_state, 0);
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(DVSSTRIDE(pipe), plane_state->color_plane[0].stride);
 	I915_WRITE_FW(DVSPOS(pipe), (crtc_y << 16) | crtc_x);
@@ -1275,7 +1275,7 @@ g4x_update_plane(struct intel_plane *plane,
 	I915_WRITE_FW(DVSSURF(pipe),
 		      intel_plane_ggtt_offset(plane_state) + dvssurf_offset);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static void
@@ -1286,14 +1286,14 @@ g4x_disable_plane(struct intel_plane *plane,
 	enum pipe pipe = plane->pipe;
 	unsigned long irqflags;
 
-	spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
+	raw_spin_lock_irqsave(&dev_priv->uncore.lock, irqflags);
 
 	I915_WRITE_FW(DVSCNTR(pipe), 0);
 	/* Disable the scaler */
 	I915_WRITE_FW(DVSSCALE(pipe), 0);
 	I915_WRITE_FW(DVSSURF(pipe), 0);
 
-	spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
+	raw_spin_unlock_irqrestore(&dev_priv->uncore.lock, irqflags);
 }
 
 static bool
diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
index d1d51e1121e2..e91502b99187 100644
--- a/drivers/gpu/drm/i915/intel_uncore.c
+++ b/drivers/gpu/drm/i915/intel_uncore.c
@@ -348,14 +348,14 @@ intel_uncore_fw_release_timer(struct hrtimer *timer)
 	if (xchg(&domain->active, false))
 		return HRTIMER_RESTART;
 
-	spin_lock_irqsave(&uncore->lock, irqflags);
+	raw_spin_lock_irqsave(&uncore->lock, irqflags);
 	if (WARN_ON(domain->wake_count == 0))
 		domain->wake_count++;
 
 	if (--domain->wake_count == 0)
 		uncore->funcs.force_wake_put(uncore, domain->mask);
 
-	spin_unlock_irqrestore(&uncore->lock, irqflags);
+	raw_spin_unlock_irqrestore(&uncore->lock, irqflags);
 
 	return HRTIMER_NORESTART;
 }
@@ -388,7 +388,7 @@ intel_uncore_forcewake_reset(struct intel_uncore *uncore)
 			intel_uncore_fw_release_timer(&domain->timer);
 		}
 
-		spin_lock_irqsave(&uncore->lock, irqflags);
+		raw_spin_lock_irqsave(&uncore->lock, irqflags);
 
 		for_each_fw_domain(domain, uncore, tmp) {
 			if (hrtimer_active(&domain->timer))
@@ -403,7 +403,7 @@ intel_uncore_forcewake_reset(struct intel_uncore *uncore)
 			break;
 		}
 
-		spin_unlock_irqrestore(&uncore->lock, irqflags);
+		raw_spin_unlock_irqrestore(&uncore->lock, irqflags);
 		cond_resched();
 	}
 
@@ -416,7 +416,7 @@ intel_uncore_forcewake_reset(struct intel_uncore *uncore)
 	fw_domains_reset(uncore, uncore->fw_domains);
 	assert_forcewakes_inactive(uncore);
 
-	spin_unlock_irqrestore(&uncore->lock, irqflags);
+	raw_spin_unlock_irqrestore(&uncore->lock, irqflags);
 
 	return fw; /* track the lost user forcewake domains */
 }
@@ -499,12 +499,12 @@ static void __intel_uncore_early_sanitize(struct intel_uncore *uncore,
 	iosf_mbi_punit_acquire();
 	intel_uncore_forcewake_reset(uncore);
 	if (restore_forcewake) {
-		spin_lock_irq(&uncore->lock);
+		raw_spin_lock_irq(&uncore->lock);
 		uncore->funcs.force_wake_get(uncore, restore_forcewake);
 
 		if (intel_uncore_has_fifo(uncore))
 			uncore->fifo_count = fifo_free_entries(uncore);
-		spin_unlock_irq(&uncore->lock);
+		raw_spin_unlock_irq(&uncore->lock);
 	}
 	iosf_mbi_punit_release();
 }
@@ -581,9 +581,9 @@ void intel_uncore_forcewake_get(struct intel_uncore *uncore,
 
 	__assert_rpm_wakelock_held(uncore->rpm);
 
-	spin_lock_irqsave(&uncore->lock, irqflags);
+	raw_spin_lock_irqsave(&uncore->lock, irqflags);
 	__intel_uncore_forcewake_get(uncore, fw_domains);
-	spin_unlock_irqrestore(&uncore->lock, irqflags);
+	raw_spin_unlock_irqrestore(&uncore->lock, irqflags);
 }
 
 /**
@@ -596,7 +596,7 @@ void intel_uncore_forcewake_get(struct intel_uncore *uncore,
  */
 void intel_uncore_forcewake_user_get(struct intel_uncore *uncore)
 {
-	spin_lock_irq(&uncore->lock);
+	raw_spin_lock_irq(&uncore->lock);
 	if (!uncore->user_forcewake.count++) {
 		intel_uncore_forcewake_get__locked(uncore, FORCEWAKE_ALL);
 
@@ -609,7 +609,7 @@ void intel_uncore_forcewake_user_get(struct intel_uncore *uncore)
 		uncore->unclaimed_mmio_check = 0;
 		i915_modparams.mmio_debug = 0;
 	}
-	spin_unlock_irq(&uncore->lock);
+	raw_spin_unlock_irq(&uncore->lock);
 }
 
 /**
@@ -621,7 +621,7 @@ void intel_uncore_forcewake_user_get(struct intel_uncore *uncore)
  */
 void intel_uncore_forcewake_user_put(struct intel_uncore *uncore)
 {
-	spin_lock_irq(&uncore->lock);
+	raw_spin_lock_irq(&uncore->lock);
 	if (!--uncore->user_forcewake.count) {
 		if (intel_uncore_unclaimed_mmio(uncore))
 			dev_info(uncore_to_i915(uncore)->drm.dev,
@@ -634,7 +634,7 @@ void intel_uncore_forcewake_user_put(struct intel_uncore *uncore)
 
 		intel_uncore_forcewake_put__locked(uncore, FORCEWAKE_ALL);
 	}
-	spin_unlock_irq(&uncore->lock);
+	raw_spin_unlock_irq(&uncore->lock);
 }
 
 /**
@@ -693,9 +693,9 @@ void intel_uncore_forcewake_put(struct intel_uncore *uncore,
 	if (!uncore->funcs.force_wake_put)
 		return;
 
-	spin_lock_irqsave(&uncore->lock, irqflags);
+	raw_spin_lock_irqsave(&uncore->lock, irqflags);
 	__intel_uncore_forcewake_put(uncore, fw_domains);
-	spin_unlock_irqrestore(&uncore->lock, irqflags);
+	raw_spin_unlock_irqrestore(&uncore->lock, irqflags);
 }
 
 /**
@@ -1093,12 +1093,12 @@ __gen2_read(64)
 	unsigned long irqflags; \
 	u##x val = 0; \
 	__assert_rpm_wakelock_held(uncore->rpm); \
-	spin_lock_irqsave(&uncore->lock, irqflags); \
+	raw_spin_lock_irqsave(&uncore->lock, irqflags); \
 	unclaimed_reg_debug(uncore, reg, true, true)
 
 #define GEN6_READ_FOOTER \
 	unclaimed_reg_debug(uncore, reg, true, false); \
-	spin_unlock_irqrestore(&uncore->lock, irqflags); \
+	raw_spin_unlock_irqrestore(&uncore->lock, irqflags); \
 	trace_i915_reg_rw(false, reg, val, sizeof(val), trace); \
 	return val
 
@@ -1205,12 +1205,12 @@ __gen2_write(32)
 	unsigned long irqflags; \
 	trace_i915_reg_rw(true, reg, val, sizeof(val), trace); \
 	__assert_rpm_wakelock_held(uncore->rpm); \
-	spin_lock_irqsave(&uncore->lock, irqflags); \
+	raw_spin_lock_irqsave(&uncore->lock, irqflags); \
 	unclaimed_reg_debug(uncore, reg, false, true)
 
 #define GEN6_WRITE_FOOTER \
 	unclaimed_reg_debug(uncore, reg, false, false); \
-	spin_unlock_irqrestore(&uncore->lock, irqflags)
+	raw_spin_unlock_irqrestore(&uncore->lock, irqflags)
 
 #define __gen6_write(x) \
 static void \
@@ -1423,11 +1423,11 @@ static void intel_uncore_fw_domains_init(struct intel_uncore *uncore)
 		fw_domain_init(uncore, FW_DOMAIN_ID_RENDER,
 			       FORCEWAKE_MT, FORCEWAKE_MT_ACK);
 
-		spin_lock_irq(&uncore->lock);
+		raw_spin_lock_irq(&uncore->lock);
 		fw_domains_get_with_thread_status(uncore, FORCEWAKE_RENDER);
 		ecobus = __raw_uncore_read32(uncore, ECOBUS);
 		fw_domains_put(uncore, FORCEWAKE_RENDER);
-		spin_unlock_irq(&uncore->lock);
+		raw_spin_unlock_irq(&uncore->lock);
 
 		if (!(ecobus & FORCEWAKE_MT_ENABLE)) {
 			DRM_INFO("No MT forcewake available on Ivybridge, this can result in issues\n");
@@ -1527,7 +1527,7 @@ static void uncore_mmio_cleanup(struct intel_uncore *uncore)
 
 void intel_uncore_init_early(struct intel_uncore *uncore)
 {
-	spin_lock_init(&uncore->lock);
+	raw_spin_lock_init(&uncore->lock);
 }
 
 int intel_uncore_init_mmio(struct intel_uncore *uncore)
@@ -1805,7 +1805,7 @@ int __intel_wait_for_register(struct intel_uncore *uncore,
 
 	might_sleep_if(slow_timeout_ms);
 
-	spin_lock_irq(&uncore->lock);
+	raw_spin_lock_irq(&uncore->lock);
 	intel_uncore_forcewake_get__locked(uncore, fw);
 
 	ret = __intel_wait_for_register_fw(uncore,
@@ -1813,7 +1813,7 @@ int __intel_wait_for_register(struct intel_uncore *uncore,
 					   fast_timeout_us, 0, &reg_value);
 
 	intel_uncore_forcewake_put__locked(uncore, fw);
-	spin_unlock_irq(&uncore->lock);
+	raw_spin_unlock_irq(&uncore->lock);
 
 	if (ret && slow_timeout_ms)
 		ret = __wait_for(reg_value = intel_uncore_read_notrace(uncore,
@@ -1840,7 +1840,7 @@ intel_uncore_arm_unclaimed_mmio_detection(struct intel_uncore *uncore)
 {
 	bool ret = false;
 
-	spin_lock_irq(&uncore->lock);
+	raw_spin_lock_irq(&uncore->lock);
 
 	if (unlikely(uncore->unclaimed_mmio_check <= 0))
 		goto out;
@@ -1857,7 +1857,7 @@ intel_uncore_arm_unclaimed_mmio_detection(struct intel_uncore *uncore)
 	}
 
 out:
-	spin_unlock_irq(&uncore->lock);
+	raw_spin_unlock_irq(&uncore->lock);
 
 	return ret;
 }
diff --git a/drivers/gpu/drm/i915/intel_uncore.h b/drivers/gpu/drm/i915/intel_uncore.h
index d6af3de70121..6518674b7871 100644
--- a/drivers/gpu/drm/i915/intel_uncore.h
+++ b/drivers/gpu/drm/i915/intel_uncore.h
@@ -99,7 +99,7 @@ struct intel_uncore {
 
 	struct i915_runtime_pm *rpm;
 
-	spinlock_t lock; /** lock is also taken in irq contexts. */
+	raw_spinlock_t lock; /** lock is also taken in irq contexts. */
 
 	unsigned int flags;
 #define UNCORE_HAS_FORCEWAKE		BIT(0)
diff --git a/drivers/gpu/drm/i915/intel_workarounds.c b/drivers/gpu/drm/i915/intel_workarounds.c
index 841b8e515f4d..47cb288f4154 100644
--- a/drivers/gpu/drm/i915/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/intel_workarounds.c
@@ -928,7 +928,7 @@ wa_list_apply(struct intel_uncore *uncore, const struct i915_wa_list *wal)
 
 	fw = wal_get_fw_for_rmw(uncore, wal);
 
-	spin_lock_irqsave(&uncore->lock, flags);
+	raw_spin_lock_irqsave(&uncore->lock, flags);
 	intel_uncore_forcewake_get__locked(uncore, fw);
 
 	for (i = 0, wa = wal->list; i < wal->count; i++, wa++) {
@@ -936,7 +936,7 @@ wa_list_apply(struct intel_uncore *uncore, const struct i915_wa_list *wal)
 	}
 
 	intel_uncore_forcewake_put__locked(uncore, fw);
-	spin_unlock_irqrestore(&uncore->lock, flags);
+	raw_spin_unlock_irqrestore(&uncore->lock, flags);
 }
 
 void intel_gt_apply_workarounds(struct drm_i915_private *i915)
-- 
2.21.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

end of thread, other threads:[~2019-09-27 15:18 UTC | newest]

Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-08-20  0:33 [PREEMPT_RT PATCH 0/3] i915 fixups for lockdep/lockdebugging Clark Williams
2019-08-20  0:33 ` [PREEMPT_RT PATCH 1/3] i915: do not call lockdep_assert_irqs_disabled() on PREEMPT_RT Clark Williams
2019-09-25 16:47   ` Sebastian Andrzej Siewior
2019-08-20  0:33 ` [PREEMPT_RT PATCH 2/3] i915: convert all irq_locks spinlocks to raw spinlocks Clark Williams
2019-09-03  8:03   ` Sebastian Andrzej Siewior
2019-09-03 13:09     ` Sebastian Andrzej Siewior
2019-09-16 21:21     ` Sean V Kelley
2019-09-27 12:29       ` Sebastian Andrzej Siewior
     [not found]         ` <CAPAFJkofsYgSOxzM4JDkOV3Ra0C-AEpU2eZJjR9F9e_qv-6=rg@mail.gmail.com>
2019-09-27 15:18           ` Sean V Kelley
2019-08-20  0:33 ` [PREEMPT_RT PATCH 3/3] i915: convert uncore lock to raw spinlock Clark Williams

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).