Merge tag 'drm-intel-next-2022-09-16-1' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

drm/i915 feature pull #2 for v6.1:

Features and functionality:
- More Meteorlake platform enabling (Radhakrishna, Imre, Madhumitha)
- Allow seamless M/N changes on eDP panels that support it (Ville)
- Switch DSC debugfs from output bpp to input bpc (Swati)

Refactoring and cleanups:
- Clocking and DPLL refactoring and cleanups to support seamless M/N (Ville)
- Plenty of VBT definition and parsing updates and cleanups (Ville)
- Extract SKL watermark code to a separate file, and clean up (Ville)
- Clean up IPC interfaces and debugfs (Jani)
- Continue moving display data under drm_i915_private display sub-struct (Jani)
- Display quirk handling refactoring and abstractions (Jani)
- Stop using implicit dev_priv in gmbus registers (Jani)
- BUG_ON() removals and conversions to drm_WARN_ON() and BUILD_BUG_ON() (Jani)
- Use drm_dp_phy_name() for logging (Jani)
- Use REG_BIT() macros for CDCLK registers (Stan)
- Move display and media IP versions to runtime info (Radhakrishna)

Fixes:
- Fix DP MST suspend to avoid use-after-free (Andrzej)
- Fix HPD suspend to avoid use-after-free for fbdev (Andrzej)
- Fix various PSR issues regarding selective update and damage clips (Jouni)
- Fix runtime pm wakerefs for driver remove and release (Mitul Golani)
- Fix conditions for filtering fixed modes for panels (Ville)
- Fix TV encoder clock computation (Ville)
- Fix dvo mode_valid hook return type (Nathan Huckleberry)

Merges:
- Backmerge drm-next to sync the DP MST atomic changes (Jani)

Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87o7vfr064.fsf@intel.com

1 files changed, 38 insertions, 1 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 53b63540c3c7..f18cc6270b2b 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1089,6 +1089,43 @@ out:
 	return err;
 }
 
+/*
+ * A single pass should suffice to release all the freed objects (along most
+ * call paths), but be a little more paranoid in that freeing the objects does
+ * take a little amount of time, during which the rcu callbacks could have added
+ * new objects into the freed list, and armed the work again.
+ */
+void i915_gem_drain_freed_objects(struct drm_i915_private *i915)
+{
+	while (atomic_read(&i915->mm.free_count)) {
+		flush_work(&i915->mm.free_work);
+		flush_delayed_work(&i915->bdev.wq);
+		rcu_barrier();
+	}
+}
+
+/*
+ * Similar to objects above (see i915_gem_drain_freed-objects), in general we
+ * have workers that are armed by RCU and then rearm themselves in their
+ * callbacks. To be paranoid, we need to drain the workqueue a second time after
+ * waiting for the RCU grace period so that we catch work queued via RCU from
+ * the first pass. As neither drain_workqueue() nor flush_workqueue() report a
+ * result, we make an assumption that we only don't require more than 3 passes
+ * to catch all _recursive_ RCU delayed work.
+ */
+void i915_gem_drain_workqueue(struct drm_i915_private *i915)
+{
+	int i;
+
+	for (i = 0; i < 3; i++) {
+		flush_workqueue(i915->wq);
+		rcu_barrier();
+		i915_gem_drain_freed_objects(i915);
+	}
+
+	drain_workqueue(i915->wq);
+}
+
 int i915_gem_init(struct drm_i915_private *dev_priv)
 {
 	int ret;
@@ -1216,7 +1253,7 @@ void i915_gem_init_early(struct drm_i915_private *dev_priv)
 	i915_gem_init__mm(dev_priv);
 	i915_gem_init__contexts(dev_priv);
 
-	spin_lock_init(&dev_priv->fb_tracking.lock);
+	spin_lock_init(&dev_priv->display.fb_tracking.lock);
 }
 
 void i915_gem_cleanup_early(struct drm_i915_private *dev_priv)