[Intel-gfx] [RFC 1/3] drm/i915: split out uncore_mmio_debug

Daniele Ceraolo Spurio daniele.ceraolospurio at intel.com
Thu Aug 8 16:43:00 UTC 2019



On 8/8/19 6:08 AM, Mika Kuoppala wrote:
> Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com> writes:
> 
>> Multiple uncore structures will share the debug infrastructure, so
>> move it to a common place and add extra locking around it.
>> Also, since we now have a separate object, it is cleaner to have
>> dedicated functions working on the object to stop and restart the
>> mmio debug. Apart from the cosmetic changes, this patch introduces
>> 2 functional updates:
>>
>> - All calls to check_for_unclaimed_mmio will now return false when
>>    the debug is suspended, not just the ones that are active only when
>>    i915_modparams.mmio_debug is set. If we don't trust the result of the
>>    check while a user is doing mmio access then we shouldn't attempt the
>>    check anywhere.
>>
>> - i915_modparams.mmio_debug is not save/restored anymore around user
>>    access. The value is now never touched by the kernel while debug is
>>    disabled so no need for save/restore.
>>
>> v2: squash mmio_debug patches, restrict mmio_debug lock usage (Chris)
>>
>> Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com>
>> Cc: Chris Wilson <chris at chris-wilson.co.uk>
>> ---
>>   drivers/gpu/drm/i915/i915_debugfs.c |  2 +-
>>   drivers/gpu/drm/i915/i915_drv.c     |  3 +-
>>   drivers/gpu/drm/i915/i915_drv.h     |  1 +
>>   drivers/gpu/drm/i915/intel_uncore.c | 91 ++++++++++++++++++++---------
>>   drivers/gpu/drm/i915/intel_uncore.h | 18 +++---
>>   5 files changed, 79 insertions(+), 36 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
>> index 3b15266c54fd..2310512111ab 100644
>> --- a/drivers/gpu/drm/i915/i915_debugfs.c
>> +++ b/drivers/gpu/drm/i915/i915_debugfs.c
>> @@ -1129,7 +1129,7 @@ static int i915_forcewake_domains(struct seq_file *m, void *data)
>>   	unsigned int tmp;
>>   
>>   	seq_printf(m, "user.bypass_count = %u\n",
>> -		   uncore->user_forcewake.count);
>> +		   uncore->user_forcewake_count);
>>   
>>   	for_each_fw_domain(fw_domain, uncore, tmp)
>>   		seq_printf(m, "%s.wake_count = %u\n",
>> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
>> index 3480db36b63f..fbbff4a133ba 100644
>> --- a/drivers/gpu/drm/i915/i915_drv.c
>> +++ b/drivers/gpu/drm/i915/i915_drv.c
>> @@ -744,6 +744,7 @@ static int i915_driver_early_probe(struct drm_i915_private *dev_priv)
>>   
>>   	intel_device_info_subplatform_init(dev_priv);
>>   
>> +	intel_uncore_mmio_debug_init_early(&dev_priv->mmio_debug);
>>   	intel_uncore_init_early(&dev_priv->uncore, dev_priv);
>>   
>>   	spin_lock_init(&dev_priv->irq_lock);
>> @@ -2044,7 +2045,7 @@ static int i915_drm_suspend_late(struct drm_device *dev, bool hibernation)
>>   
>>   out:
>>   	enable_rpm_wakeref_asserts(rpm);
>> -	if (!dev_priv->uncore.user_forcewake.count)
>> +	if (!dev_priv->uncore.user_forcewake_count)
>>   		intel_runtime_pm_driver_release(rpm);
>>   
>>   	return ret;
>> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
>> index c9476f24f5c1..13c27a75dca8 100644
>> --- a/drivers/gpu/drm/i915/i915_drv.h
>> +++ b/drivers/gpu/drm/i915/i915_drv.h
>> @@ -1346,6 +1346,7 @@ struct drm_i915_private {
>>   	resource_size_t stolen_usable_size;	/* Total size minus reserved ranges */
>>   
>>   	struct intel_uncore uncore;
>> +	struct intel_uncore_mmio_debug mmio_debug;
>>   
>>   	struct i915_virtual_gpu vgpu;
>>   
>> diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
>> index 39e8710afdd9..9e583f13a9e4 100644
>> --- a/drivers/gpu/drm/i915/intel_uncore.c
>> +++ b/drivers/gpu/drm/i915/intel_uncore.c
>> @@ -34,6 +34,32 @@
>>   
>>   #define __raw_posting_read(...) ((void)__raw_uncore_read32(__VA_ARGS__))
>>   
>> +void
>> +intel_uncore_mmio_debug_init_early(struct intel_uncore_mmio_debug *mmio_debug)
>> +{
>> +	spin_lock_init(&mmio_debug->lock);
>> +	mmio_debug->unclaimed_mmio_check = 1;
>> +}
>> +
>> +static void mmio_debug_suspend(struct intel_uncore_mmio_debug *mmio_debug)
>> +{
>> +	lockdep_assert_held(&mmio_debug->lock);
>> +
>> +	/* Save and disable mmio debugging for the user bypass */
>> +	if (!mmio_debug->suspend_count++) {
>> +		mmio_debug->saved_mmio_check = mmio_debug->unclaimed_mmio_check;
>> +		mmio_debug->unclaimed_mmio_check = 0;
>> +	}
>> +}
>> +
>> +static void mmio_debug_resume(struct intel_uncore_mmio_debug *mmio_debug)
>> +{
>> +	lockdep_assert_held(&mmio_debug->lock);
>> +
>> +	if (!--mmio_debug->suspend_count)
>> +		mmio_debug->unclaimed_mmio_check = mmio_debug->saved_mmio_check;
>> +}
>> +
>>   static const char * const forcewake_domain_names[] = {
>>   	"render",
>>   	"blitter",
>> @@ -476,6 +502,11 @@ check_for_unclaimed_mmio(struct intel_uncore *uncore)
>>   {
>>   	bool ret = false;
>>   
>> +	lockdep_assert_held(&uncore->debug->lock);
>> +
>> +	if (uncore->debug->suspend_count)
>> +		return false;
>> +
>>   	if (intel_uncore_has_fpga_dbg_unclaimed(uncore))
>>   		ret |= fpga_check_for_unclaimed_mmio(uncore);
>>   
>> @@ -608,17 +639,11 @@ void intel_uncore_forcewake_get(struct intel_uncore *uncore,
>>   void intel_uncore_forcewake_user_get(struct intel_uncore *uncore)
>>   {
>>   	spin_lock_irq(&uncore->lock);
>> -	if (!uncore->user_forcewake.count++) {
>> +	if (!uncore->user_forcewake_count++) {
>>   		intel_uncore_forcewake_get__locked(uncore, FORCEWAKE_ALL);
>> -
>> -		/* Save and disable mmio debugging for the user bypass */
>> -		uncore->user_forcewake.saved_mmio_check =
>> -			uncore->unclaimed_mmio_check;
>> -		uncore->user_forcewake.saved_mmio_debug =
>> -			i915_modparams.mmio_debug;
>> -
>> -		uncore->unclaimed_mmio_check = 0;
>> -		i915_modparams.mmio_debug = 0;
>> +		spin_lock(&uncore->debug->lock);
> 
> For me it looks like you need spin_lock_irq here also
> like with the parent lock above.

The parent lock should ensure that the irqs are already off at this 
point, so no need to disable them twice. Also, spin_unlock_irq can't be 
called twice because the first of the 2 calls would re-enable interrupts 
while the second lock is still taken. Irqflags would solve that, but 
still wouldn't give any benefit compared to a straight lock IMO.

Daniele

> -Mika
> 
>> +		mmio_debug_suspend(uncore->debug);
>> +		spin_unlock(&uncore->debug->lock);
>>   	}
>>   	spin_unlock_irq(&uncore->lock);
>>   }
>> @@ -633,15 +658,14 @@ 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);
>> -	if (!--uncore->user_forcewake.count) {
>> -		if (intel_uncore_unclaimed_mmio(uncore))
>> +	if (!--uncore->user_forcewake_count) {
>> +		spin_lock(&uncore->debug->lock);
>> +		mmio_debug_resume(uncore->debug);
>> +
>> +		if (check_for_unclaimed_mmio(uncore))
>>   			dev_info(uncore->i915->drm.dev,
>>   				 "Invalid mmio detected during user access\n");
>> -
>> -		uncore->unclaimed_mmio_check =
>> -			uncore->user_forcewake.saved_mmio_check;
>> -		i915_modparams.mmio_debug =
>> -			uncore->user_forcewake.saved_mmio_debug;
>> +		spin_unlock(&uncore->debug->lock);
>>   
>>   		intel_uncore_forcewake_put__locked(uncore, FORCEWAKE_ALL);
>>   	}
>> @@ -1088,7 +1112,16 @@ unclaimed_reg_debug(struct intel_uncore *uncore,
>>   	if (likely(!i915_modparams.mmio_debug))
>>   		return;
>>   
>> +	/* interrupts are disabled and re-enabled around uncore->lock usage */
>> +	lockdep_assert_held(&uncore->lock);
>> +
>> +	if (before)
>> +		spin_lock(&uncore->debug->lock);
>> +
>>   	__unclaimed_reg_debug(uncore, reg, read, before);
>> +
>> +	if (!before)
>> +		spin_unlock(&uncore->debug->lock);
>>   }
>>   
>>   #define GEN2_READ_HEADER(x) \
>> @@ -1607,6 +1640,7 @@ void intel_uncore_init_early(struct intel_uncore *uncore,
>>   	spin_lock_init(&uncore->lock);
>>   	uncore->i915 = i915;
>>   	uncore->rpm = &i915->runtime_pm;
>> +	uncore->debug = &i915->mmio_debug;
>>   }
>>   
>>   static void uncore_raw_init(struct intel_uncore *uncore)
>> @@ -1632,7 +1666,6 @@ static int uncore_forcewake_init(struct intel_uncore *uncore)
>>   	ret = intel_uncore_fw_domains_init(uncore);
>>   	if (ret)
>>   		return ret;
>> -
>>   	forcewake_early_sanitize(uncore, 0);
>>   
>>   	if (IS_GEN_RANGE(i915, 6, 7)) {
>> @@ -1681,8 +1714,6 @@ int intel_uncore_init_mmio(struct intel_uncore *uncore)
>>   	if (INTEL_GEN(i915) > 5 && !intel_vgpu_active(i915))
>>   		uncore->flags |= UNCORE_HAS_FORCEWAKE;
>>   
>> -	uncore->unclaimed_mmio_check = 1;
>> -
>>   	if (!intel_uncore_has_forcewake(uncore)) {
>>   		uncore_raw_init(uncore);
>>   	} else {
>> @@ -1707,7 +1738,7 @@ int intel_uncore_init_mmio(struct intel_uncore *uncore)
>>   		uncore->flags |= UNCORE_HAS_FIFO;
>>   
>>   	/* clear out unclaimed reg detection bit */
>> -	if (check_for_unclaimed_mmio(uncore))
>> +	if (intel_uncore_unclaimed_mmio(uncore))
>>   		DRM_DEBUG("unclaimed mmio detected on uncore init, clearing\n");
>>   
>>   	return 0;
>> @@ -1952,7 +1983,13 @@ int __intel_wait_for_register(struct intel_uncore *uncore,
>>   
>>   bool intel_uncore_unclaimed_mmio(struct intel_uncore *uncore)
>>   {
>> -	return check_for_unclaimed_mmio(uncore);
>> +	bool ret;
>> +
>> +	spin_lock_irq(&uncore->debug->lock);
>> +	ret = check_for_unclaimed_mmio(uncore);
>> +	spin_unlock_irq(&uncore->debug->lock);
>> +
>> +	return ret;
>>   }
>>   
>>   bool
>> @@ -1960,24 +1997,24 @@ intel_uncore_arm_unclaimed_mmio_detection(struct intel_uncore *uncore)
>>   {
>>   	bool ret = false;
>>   
>> -	spin_lock_irq(&uncore->lock);
>> +	spin_lock_irq(&uncore->debug->lock);
>>   
>> -	if (unlikely(uncore->unclaimed_mmio_check <= 0))
>> +	if (unlikely(uncore->debug->unclaimed_mmio_check <= 0))
>>   		goto out;
>>   
>> -	if (unlikely(intel_uncore_unclaimed_mmio(uncore))) {
>> +	if (unlikely(check_for_unclaimed_mmio(uncore))) {
>>   		if (!i915_modparams.mmio_debug) {
>>   			DRM_DEBUG("Unclaimed register detected, "
>>   				  "enabling oneshot unclaimed register reporting. "
>>   				  "Please use i915.mmio_debug=N for more information.\n");
>>   			i915_modparams.mmio_debug++;
>>   		}
>> -		uncore->unclaimed_mmio_check--;
>> +		uncore->debug->unclaimed_mmio_check--;
>>   		ret = true;
>>   	}
>>   
>>   out:
>> -	spin_unlock_irq(&uncore->lock);
>> +	spin_unlock_irq(&uncore->debug->lock);
>>   
>>   	return ret;
>>   }
>> diff --git a/drivers/gpu/drm/i915/intel_uncore.h b/drivers/gpu/drm/i915/intel_uncore.h
>> index e603d210a34d..414fc2cb0459 100644
>> --- a/drivers/gpu/drm/i915/intel_uncore.h
>> +++ b/drivers/gpu/drm/i915/intel_uncore.h
>> @@ -36,6 +36,13 @@ struct drm_i915_private;
>>   struct intel_runtime_pm;
>>   struct intel_uncore;
>>   
>> +struct intel_uncore_mmio_debug {
>> +	spinlock_t lock; /** lock is also taken in irq contexts. */
>> +	int unclaimed_mmio_check;
>> +	int saved_mmio_check;
>> +	u32 suspend_count;
>> +};
>> +
>>   enum forcewake_domain_id {
>>   	FW_DOMAIN_ID_RENDER = 0,
>>   	FW_DOMAIN_ID_BLITTER,
>> @@ -137,14 +144,9 @@ struct intel_uncore {
>>   		u32 __iomem *reg_ack;
>>   	} *fw_domain[FW_DOMAIN_ID_COUNT];
>>   
>> -	struct {
>> -		unsigned int count;
>> -
>> -		int saved_mmio_check;
>> -		int saved_mmio_debug;
>> -	} user_forcewake;
>> +	unsigned int user_forcewake_count;
>>   
>> -	int unclaimed_mmio_check;
>> +	struct intel_uncore_mmio_debug *debug;
>>   };
>>   
>>   /* Iterate over initialised fw domains */
>> @@ -179,6 +181,8 @@ intel_uncore_has_fifo(const struct intel_uncore *uncore)
>>   	return uncore->flags & UNCORE_HAS_FIFO;
>>   }
>>   
>> +void
>> +intel_uncore_mmio_debug_init_early(struct intel_uncore_mmio_debug *mmio_debug);
>>   void intel_uncore_init_early(struct intel_uncore *uncore,
>>   			     struct drm_i915_private *i915);
>>   int intel_uncore_init_mmio(struct intel_uncore *uncore);
>> -- 
>> 2.22.0
>>
>> _______________________________________________
>> Intel-gfx mailing list
>> Intel-gfx at lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/intel-gfx


More information about the Intel-gfx mailing list