Merge tag 'drm-next-2021-04-30' of git://anongit.freedesktop.org/drm/drm

Pull more drm updates from Dave Airlie:
 "Looks like I missed a tegra feature request for next, but should still
  be fine since it's pretty self contained.

  Apart from that got a set of i915 and amdgpu fixes as per usual along
  with a few misc fixes.

  tegra:
   - Tegra186 hardware cursor support
   - better capability reporting for different SoC
   - better framebuffer modifier support
   - host1x fixes

  ttm:
   - fix unswappable BO handling

  efifb:
   - check for PCI before using it

  amdgpu:
   - Fixes for Aldebaran
   - Display LTTPR fixes
   - eDP fixes
   - Fixes for Vangogh
   - RAS fixes
   - ASPM support
   - Renoir SMU fixes
   - Modifier fixes
   - Misc code cleanups
   - Freesync fixes

  i915:
   - Several fixes to GLK handling in recent display refactoring
   - Rare watchdog timer race fix
   - Cppcheck redundant condition fix
   - Overlay error code propagation fix
   - Documentation fix
   - gvt: Remove one unused function warning
   - gvt: Fix intel_gvt_init_device() return type
   - gvt: Remove one duplicated register accessible check"

* tag 'drm-next-2021-04-30' of git://anongit.freedesktop.org/drm/drm: (111 commits)
  efifb: Check efifb_pci_dev before using it
  drm/i915: Fix docbook descriptions for i915_gem_shrinker
  drm/i915: fix an error code in intel_overlay_do_put_image()
  drm/i915/display/psr: Fix cppcheck warnings
  drm/i915: Disable LTTPR detection on GLK once again
  drm/i915: Restore lost glk ccs w/a
  drm/i915: Restore lost glk FBC 16bpp w/a
  drm/i915: Take request reference before arming the watchdog timer
  drm/ttm: fix error handling if no BO can be swapped out v4
  drm/i915/gvt: Remove duplicated register accessible check
  drm/amdgpu/gmc9: remove dummy read workaround for newer chips
  drm/amdgpu: Add mem sync flag for IB allocated by SA
  drm/amdgpu: Fix SDMA RAS error reporting on Aldebaran
  drm/amdgpu: Reset RAS error count and status regs
  Revert "drm/amdgpu: workaround the TMR MC address issue (v2)"
  drm/amd/display: 3.2.132
  drm/amd/display: [FW Promotion] Release 0.0.62
  drm/amd/display: add helper for enabling mst stream features
  drm/amd/display: Report Proper Quantization Range in AVI Infoframe
  drm/amd/display: Fix call to pass bpp in 16ths of a bit
  ...

14 files changed, 218 insertions, 116 deletions

diff --git a/drivers/gpu/host1x/bus.c b/drivers/gpu/host1x/bus.c
index 68a766ff0e9d..46f69c532b6b 100644
--- a/drivers/gpu/host1x/bus.c
+++ b/drivers/gpu/host1x/bus.c
@@ -197,6 +197,17 @@ int host1x_device_init(struct host1x_device *device)
 	mutex_lock(&device->clients_lock);
 
 	list_for_each_entry(client, &device->clients, list) {
+		if (client->ops && client->ops->early_init) {
+			err = client->ops->early_init(client);
+			if (err < 0) {
+				dev_err(&device->dev, "failed to early initialize %s: %d\n",
+					dev_name(client->dev), err);
+				goto teardown_late;
+			}
+		}
+	}
+
+	list_for_each_entry(client, &device->clients, list) {
 		if (client->ops && client->ops->init) {
 			err = client->ops->init(client);
 			if (err < 0) {
@@ -217,6 +228,14 @@ teardown:
 		if (client->ops->exit)
 			client->ops->exit(client);
 
+	/* reset client to end of list for late teardown */
+	client = list_entry(&device->clients, struct host1x_client, list);
+
+teardown_late:
+	list_for_each_entry_continue_reverse(client, &device->clients, list)
+		if (client->ops->late_exit)
+			client->ops->late_exit(client);
+
 	mutex_unlock(&device->clients_lock);
 	return err;
 }
@@ -251,6 +270,18 @@ int host1x_device_exit(struct host1x_device *device)
 		}
 	}
 
+	list_for_each_entry_reverse(client, &device->clients, list) {
+		if (client->ops && client->ops->late_exit) {
+			err = client->ops->late_exit(client);
+			if (err < 0) {
+				dev_err(&device->dev, "failed to late cleanup %s: %d\n",
+					dev_name(client->dev), err);
+				mutex_unlock(&device->clients_lock);
+				return err;
+			}
+		}
+	}
+
 	mutex_unlock(&device->clients_lock);
 
 	return 0;
diff --git a/drivers/gpu/host1x/cdma.c b/drivers/gpu/host1x/cdma.c
index e8d3fda91d8a..6e6ca774f68d 100644
--- a/drivers/gpu/host1x/cdma.c
+++ b/drivers/gpu/host1x/cdma.c
@@ -273,15 +273,13 @@ static int host1x_cdma_wait_pushbuffer_space(struct host1x *host1x,
 static void cdma_start_timer_locked(struct host1x_cdma *cdma,
 				    struct host1x_job *job)
 {
-	struct host1x *host = cdma_to_host1x(cdma);
-
 	if (cdma->timeout.client) {
 		/* timer already started */
 		return;
 	}
 
 	cdma->timeout.client = job->client;
-	cdma->timeout.syncpt = host1x_syncpt_get(host, job->syncpt_id);
+	cdma->timeout.syncpt = job->syncpt;
 	cdma->timeout.syncpt_val = job->syncpt_end;
 	cdma->timeout.start_ktime = ktime_get();
 
@@ -312,7 +310,6 @@ static void stop_cdma_timer_locked(struct host1x_cdma *cdma)
 static void update_cdma_locked(struct host1x_cdma *cdma)
 {
 	bool signal = false;
-	struct host1x *host1x = cdma_to_host1x(cdma);
 	struct host1x_job *job, *n;
 
 	/* If CDMA is stopped, queue is cleared and we can return */
@@ -324,8 +321,7 @@ static void update_cdma_locked(struct host1x_cdma *cdma)
 	 * to consume as many sync queue entries as possible without blocking
 	 */
 	list_for_each_entry_safe(job, n, &cdma->sync_queue, list) {
-		struct host1x_syncpt *sp =
-			host1x_syncpt_get(host1x, job->syncpt_id);
+		struct host1x_syncpt *sp = job->syncpt;
 
 		/* Check whether this syncpt has completed, and bail if not */
 		if (!host1x_syncpt_is_expired(sp, job->syncpt_end)) {
@@ -499,8 +495,7 @@ int host1x_cdma_begin(struct host1x_cdma *cdma, struct host1x_job *job)
 		if (!cdma->timeout.initialized) {
 			int err;
 
-			err = host1x_hw_cdma_timeout_init(host1x, cdma,
-							  job->syncpt_id);
+			err = host1x_hw_cdma_timeout_init(host1x, cdma);
 			if (err) {
 				mutex_unlock(&cdma->lock);
 				return err;
diff --git a/drivers/gpu/host1x/debug.c b/drivers/gpu/host1x/debug.c
index 1b4997bda1c7..8a14880c61bb 100644
--- a/drivers/gpu/host1x/debug.c
+++ b/drivers/gpu/host1x/debug.c
@@ -69,6 +69,7 @@ static int show_channel(struct host1x_channel *ch, void *data, bool show_fifo)
 
 static void show_syncpts(struct host1x *m, struct output *o)
 {
+	struct list_head *pos;
 	unsigned int i;
 
 	host1x_debug_output(o, "---- syncpts ----\n");
@@ -76,12 +77,19 @@ static void show_syncpts(struct host1x *m, struct output *o)
 	for (i = 0; i < host1x_syncpt_nb_pts(m); i++) {
 		u32 max = host1x_syncpt_read_max(m->syncpt + i);
 		u32 min = host1x_syncpt_load(m->syncpt + i);
+		unsigned int waiters = 0;
 
-		if (!min && !max)
+		spin_lock(&m->syncpt[i].intr.lock);
+		list_for_each(pos, &m->syncpt[i].intr.wait_head)
+			waiters++;
+		spin_unlock(&m->syncpt[i].intr.lock);
+
+		if (!min && !max && !waiters)
 			continue;
 
-		host1x_debug_output(o, "id %u (%s) min %d max %d\n",
-				    i, m->syncpt[i].name, min, max);
+		host1x_debug_output(o,
+				    "id %u (%s) min %d max %d (%d waiters)\n",
+				    i, m->syncpt[i].name, min, max, waiters);
 	}
 
 	for (i = 0; i < host1x_syncpt_nb_bases(m); i++) {
diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c
index d0ebb70e2fdd..fbb6447b8659 100644
--- a/drivers/gpu/host1x/dev.c
+++ b/drivers/gpu/host1x/dev.c
@@ -77,6 +77,7 @@ static const struct host1x_info host1x01_info = {
 	.has_hypervisor = false,
 	.num_sid_entries = 0,
 	.sid_table = NULL,
+	.reserve_vblank_syncpts = true,
 };
 
 static const struct host1x_info host1x02_info = {
@@ -91,6 +92,7 @@ static const struct host1x_info host1x02_info = {
 	.has_hypervisor = false,
 	.num_sid_entries = 0,
 	.sid_table = NULL,
+	.reserve_vblank_syncpts = true,
 };
 
 static const struct host1x_info host1x04_info = {
@@ -105,6 +107,7 @@ static const struct host1x_info host1x04_info = {
 	.has_hypervisor = false,
 	.num_sid_entries = 0,
 	.sid_table = NULL,
+	.reserve_vblank_syncpts = false,
 };
 
 static const struct host1x_info host1x05_info = {
@@ -119,6 +122,7 @@ static const struct host1x_info host1x05_info = {
 	.has_hypervisor = false,
 	.num_sid_entries = 0,
 	.sid_table = NULL,
+	.reserve_vblank_syncpts = false,
 };
 
 static const struct host1x_sid_entry tegra186_sid_table[] = {
@@ -142,6 +146,7 @@ static const struct host1x_info host1x06_info = {
 	.has_hypervisor = true,
 	.num_sid_entries = ARRAY_SIZE(tegra186_sid_table),
 	.sid_table = tegra186_sid_table,
+	.reserve_vblank_syncpts = false,
 };
 
 static const struct host1x_sid_entry tegra194_sid_table[] = {
@@ -165,6 +170,7 @@ static const struct host1x_info host1x07_info = {
 	.has_hypervisor = true,
 	.num_sid_entries = ARRAY_SIZE(tegra194_sid_table),
 	.sid_table = tegra194_sid_table,
+	.reserve_vblank_syncpts = false,
 };
 
 static const struct of_device_id host1x_of_match[] = {
diff --git a/drivers/gpu/host1x/dev.h b/drivers/gpu/host1x/dev.h
index f781a9b0f39d..fa6d4bc46e98 100644
--- a/drivers/gpu/host1x/dev.h
+++ b/drivers/gpu/host1x/dev.h
@@ -37,7 +37,7 @@ struct host1x_cdma_ops {
 	void (*start)(struct host1x_cdma *cdma);
 	void (*stop)(struct host1x_cdma *cdma);
 	void (*flush)(struct  host1x_cdma *cdma);
-	int (*timeout_init)(struct host1x_cdma *cdma, unsigned int syncpt);
+	int (*timeout_init)(struct host1x_cdma *cdma);
 	void (*timeout_destroy)(struct host1x_cdma *cdma);
 	void (*freeze)(struct host1x_cdma *cdma);
 	void (*resume)(struct host1x_cdma *cdma, u32 getptr);
@@ -101,6 +101,12 @@ struct host1x_info {
 	bool has_hypervisor; /* has hypervisor registers */
 	unsigned int num_sid_entries;
 	const struct host1x_sid_entry *sid_table;
+	/*
+	 * On T20-T148, the boot chain may setup DC to increment syncpoints
+	 * 26/27 on VBLANK. As such we cannot use these syncpoints until
+	 * the display driver disables VBLANK increments.
+	 */
+	bool reserve_vblank_syncpts;
 };
 
 struct host1x {
@@ -261,10 +267,9 @@ static inline void host1x_hw_cdma_flush(struct host1x *host,
 }
 
 static inline int host1x_hw_cdma_timeout_init(struct host1x *host,
-					      struct host1x_cdma *cdma,
-					      unsigned int syncpt)
+					      struct host1x_cdma *cdma)
 {
-	return host->cdma_op->timeout_init(cdma, syncpt);
+	return host->cdma_op->timeout_init(cdma);
 }
 
 static inline void host1x_hw_cdma_timeout_destroy(struct host1x *host,
diff --git a/drivers/gpu/host1x/hw/cdma_hw.c b/drivers/gpu/host1x/hw/cdma_hw.c
index 2f3bf94cf365..e49cd5b8f735 100644
--- a/drivers/gpu/host1x/hw/cdma_hw.c
+++ b/drivers/gpu/host1x/hw/cdma_hw.c
@@ -295,7 +295,7 @@ static void cdma_timeout_handler(struct work_struct *work)
 /*
  * Init timeout resources
  */
-static int cdma_timeout_init(struct host1x_cdma *cdma, unsigned int syncpt)
+static int cdma_timeout_init(struct host1x_cdma *cdma)
 {
 	INIT_DELAYED_WORK(&cdma->timeout.wq, cdma_timeout_handler);
 	cdma->timeout.initialized = true;
diff --git a/drivers/gpu/host1x/hw/channel_hw.c b/drivers/gpu/host1x/hw/channel_hw.c
index 5eaa29d171c9..d4c28faf27d1 100644
--- a/drivers/gpu/host1x/hw/channel_hw.c
+++ b/drivers/gpu/host1x/hw/channel_hw.c
@@ -86,8 +86,7 @@ static void submit_gathers(struct host1x_job *job)
 
 static inline void synchronize_syncpt_base(struct host1x_job *job)
 {
-	struct host1x *host = dev_get_drvdata(job->channel->dev->parent);
-	struct host1x_syncpt *sp = host->syncpt + job->syncpt_id;
+	struct host1x_syncpt *sp = job->syncpt;
 	unsigned int id;
 	u32 value;
 
@@ -118,7 +117,7 @@ static void host1x_channel_set_streamid(struct host1x_channel *channel)
 static int channel_submit(struct host1x_job *job)
 {
 	struct host1x_channel *ch = job->channel;
-	struct host1x_syncpt *sp;
+	struct host1x_syncpt *sp = job->syncpt;
 	u32 user_syncpt_incrs = job->syncpt_incrs;
 	u32 prev_max = 0;
 	u32 syncval;
@@ -126,10 +125,9 @@ static int channel_submit(struct host1x_job *job)
 	struct host1x_waitlist *completed_waiter = NULL;
 	struct host1x *host = dev_get_drvdata(ch->dev->parent);
 
-	sp = host->syncpt + job->syncpt_id;
 	trace_host1x_channel_submit(dev_name(ch->dev),
 				    job->num_gathers, job->num_relocs,
-				    job->syncpt_id, job->syncpt_incrs);
+				    job->syncpt->id, job->syncpt_incrs);
 
 	/* before error checks, return current max */
 	prev_max = job->syncpt_end = host1x_syncpt_read_max(sp);
@@ -163,7 +161,7 @@ static int channel_submit(struct host1x_job *job)
 		host1x_cdma_push(&ch->cdma,
 				 host1x_opcode_setclass(HOST1X_CLASS_HOST1X,
 					host1x_uclass_wait_syncpt_r(), 1),
-				 host1x_class_host_wait_syncpt(job->syncpt_id,
+				 host1x_class_host_wait_syncpt(job->syncpt->id,
 					host1x_syncpt_read_max(sp)));
 	}
 
diff --git a/drivers/gpu/host1x/hw/debug_hw.c b/drivers/gpu/host1x/hw/debug_hw.c
index f31bcfa1b837..ceb48229d14b 100644
--- a/drivers/gpu/host1x/hw/debug_hw.c
+++ b/drivers/gpu/host1x/hw/debug_hw.c
@@ -204,7 +204,7 @@ static void show_channel_gathers(struct output *o, struct host1x_cdma *cdma)
 		unsigned int i;
 
 		host1x_debug_output(o, "\n%p: JOB, syncpt_id=%d, syncpt_val=%d, first_get=%08x, timeout=%d num_slots=%d, num_handles=%d\n",
-				    job, job->syncpt_id, job->syncpt_end,
+				    job, job->syncpt->id, job->syncpt_end,
 				    job->first_get, job->timeout,
 				    job->num_slots, job->num_unpins);
 
diff --git a/drivers/gpu/host1x/hw/hw_host1x07_vm.h b/drivers/gpu/host1x/hw/hw_host1x07_vm.h
index 3058b3c9a91d..b766851d5b83 100644
--- a/drivers/gpu/host1x/hw/hw_host1x07_vm.h
+++ b/drivers/gpu/host1x/hw/hw_host1x07_vm.h
@@ -29,6 +29,6 @@
 #define HOST1X_SYNC_SYNCPT_THRESH_INT_ENABLE_CPU0(x)	(0x652c + 4 * (x))
 #define HOST1X_SYNC_SYNCPT_THRESH_INT_DISABLE(x)	(0x6590 + 4 * (x))
 #define HOST1X_SYNC_SYNCPT(x)				(0x8080 + 4 * (x))
-#define HOST1X_SYNC_SYNCPT_INT_THRESH(x)		(0x8d00 + 4 * (x))
+#define HOST1X_SYNC_SYNCPT_INT_THRESH(x)		(0x9980 + 4 * (x))
 #define HOST1X_SYNC_SYNCPT_CH_APP(x)			(0xa604 + 4 * (x))
 #define HOST1X_SYNC_SYNCPT_CH_APP_CH(v)			(((v) & 0x3f) << 8)
diff --git a/drivers/gpu/host1x/intr.c b/drivers/gpu/host1x/intr.c
index 9245add23b5d..6d1f3c0fdbe7 100644
--- a/drivers/gpu/host1x/intr.c
+++ b/drivers/gpu/host1x/intr.c
@@ -235,25 +235,37 @@ int host1x_intr_add_action(struct host1x *host, struct host1x_syncpt *syncpt,
 			host1x_hw_intr_enable_syncpt_intr(host, syncpt->id);
 	}
 
-	spin_unlock(&syncpt->intr.lock);
-
 	if (ref)
 		*ref = waiter;
+
+	spin_unlock(&syncpt->intr.lock);
+
 	return 0;
 }
 
-void host1x_intr_put_ref(struct host1x *host, unsigned int id, void *ref)
+void host1x_intr_put_ref(struct host1x *host, unsigned int id, void *ref,
+			 bool flush)
 {
 	struct host1x_waitlist *waiter = ref;
 	struct host1x_syncpt *syncpt;
 
-	while (atomic_cmpxchg(&waiter->state, WLS_PENDING, WLS_CANCELLED) ==
-	       WLS_REMOVED)
-		schedule();
+	atomic_cmpxchg(&waiter->state, WLS_PENDING, WLS_CANCELLED);
 
 	syncpt = host->syncpt + id;
-	(void)process_wait_list(host, syncpt,
-				host1x_syncpt_load(host->syncpt + id));
+
+	spin_lock(&syncpt->intr.lock);
+	if (atomic_cmpxchg(&waiter->state, WLS_CANCELLED, WLS_HANDLED) ==
+	    WLS_CANCELLED) {
+		list_del(&waiter->list);
+		kref_put(&waiter->refcount, waiter_release);
+	}
+	spin_unlock(&syncpt->intr.lock);
+
+	if (flush) {
+		/* Wait until any concurrently executing handler has finished. */
+		while (atomic_read(&waiter->state) != WLS_HANDLED)
+			schedule();
+	}
 
 	kref_put(&waiter->refcount, waiter_release);
 }
diff --git a/drivers/gpu/host1x/intr.h b/drivers/gpu/host1x/intr.h
index aac38194398f..6ea55e615e3a 100644
--- a/drivers/gpu/host1x/intr.h
+++ b/drivers/gpu/host1x/intr.h
@@ -74,8 +74,10 @@ int host1x_intr_add_action(struct host1x *host, struct host1x_syncpt *syncpt,
  * Unreference an action submitted to host1x_intr_add_action().
  * You must call this if you passed non-NULL as ref.
  * @ref the ref returned from host1x_intr_add_action()
+ * @flush wait until any pending handlers have completed before returning.
  */
-void host1x_intr_put_ref(struct host1x *host, unsigned int id, void *ref);
+void host1x_intr_put_ref(struct host1x *host, unsigned int id, void *ref,
+			 bool flush);
 
 /* Initialize host1x sync point interrupt */
 int host1x_intr_init(struct host1x *host, unsigned int irq_sync);
diff --git a/drivers/gpu/host1x/job.c b/drivers/gpu/host1x/job.c
index 82d0a60ba3f7..adbdc225de8d 100644
--- a/drivers/gpu/host1x/job.c
+++ b/drivers/gpu/host1x/job.c
@@ -79,6 +79,9 @@ static void job_free(struct kref *ref)
 {
 	struct host1x_job *job = container_of(ref, struct host1x_job, ref);
 
+	if (job->syncpt)
+		host1x_syncpt_put(job->syncpt);
+
 	kfree(job);
 }
 
@@ -674,7 +677,7 @@ EXPORT_SYMBOL(host1x_job_unpin);
  */
 void host1x_job_dump(struct device *dev, struct host1x_job *job)
 {
-	dev_dbg(dev, "    SYNCPT_ID   %d\n", job->syncpt_id);
+	dev_dbg(dev, "    SYNCPT_ID   %d\n", job->syncpt->id);
 	dev_dbg(dev, "    SYNCPT_VAL  %d\n", job->syncpt_end);
 	dev_dbg(dev, "    FIRST_GET   0x%x\n", job->first_get);
 	dev_dbg(dev, "    TIMEOUT     %d\n", job->timeout);
diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c
index fce7892d5137..e648ebbb2027 100644
--- a/drivers/gpu/host1x/syncpt.c
+++ b/drivers/gpu/host1x/syncpt.c
@@ -42,17 +42,32 @@ static void host1x_syncpt_base_free(struct host1x_syncpt_base *base)
 		base->requested = false;
 }
 
-static struct host1x_syncpt *host1x_syncpt_alloc(struct host1x *host,
-						 struct host1x_client *client,
-						 unsigned long flags)
+/**
+ * host1x_syncpt_alloc() - allocate a syncpoint
+ * @host: host1x device data
+ * @flags: bitfield of HOST1X_SYNCPT_* flags
+ * @name: name for the syncpoint for use in debug prints
+ *
+ * Allocates a hardware syncpoint for the caller's use. The caller then has
+ * the sole authority to mutate the syncpoint's value until it is freed again.
+ *
+ * If no free syncpoints are available, or a NULL name was specified, returns
+ * NULL.
+ */
+struct host1x_syncpt *host1x_syncpt_alloc(struct host1x *host,
+					  unsigned long flags,
+					  const char *name)
 {
 	struct host1x_syncpt *sp = host->syncpt;
+	char *full_name;
 	unsigned int i;
-	char *name;
+
+	if (!name)
+		return NULL;
 
 	mutex_lock(&host->syncpt_mutex);
 
-	for (i = 0; i < host->info->nb_pts && sp->name; i++, sp++)
+	for (i = 0; i < host->info->nb_pts && kref_read(&sp->ref); i++, sp++)
 		;
 
 	if (i >= host->info->nb_pts)
@@ -64,19 +79,19 @@ static struct host1x_syncpt *host1x_syncpt_alloc(struct host1x *host,
 			goto unlock;
 	}
 
-	name = kasprintf(GFP_KERNEL, "%02u-%s", sp->id,
-			 client ? dev_name(client->dev) : NULL);
-	if (!name)
+	full_name = kasprintf(GFP_KERNEL, "%u-%s", sp->id, name);
+	if (!full_name)
 		goto free_base;
 
-	sp->client = client;
-	sp->name = name;
+	sp->name = full_name;
 
 	if (flags & HOST1X_SYNCPT_CLIENT_MANAGED)
 		sp->client_managed = true;
 	else
 		sp->client_managed = false;
 
+	kref_init(&sp->ref);
+
 	mutex_unlock(&host->syncpt_mutex);
 	return sp;
 
@@ -87,6 +102,7 @@ unlock:
 	mutex_unlock(&host->syncpt_mutex);
 	return NULL;
 }
+EXPORT_SYMBOL(host1x_syncpt_alloc);
 
 /**
  * host1x_syncpt_id() - retrieve syncpoint ID
@@ -294,7 +310,7 @@ int host1x_syncpt_wait(struct host1x_syncpt *sp, u32 thresh, long timeout,
 		}
 	}
 
-	host1x_intr_put_ref(sp->host, sp->id, ref);
+	host1x_intr_put_ref(sp->host, sp->id, ref, true);
 
 done:
 	return err;
@@ -307,59 +323,12 @@ EXPORT_SYMBOL(host1x_syncpt_wait);
 bool host1x_syncpt_is_expired(struct host1x_syncpt *sp, u32 thresh)
 {
 	u32 current_val;
-	u32 future_val;
 
 	smp_rmb();
 
 	current_val = (u32)atomic_read(&sp->min_val);
-	future_val = (u32)atomic_read(&sp->max_val);
-
-	/* Note the use of unsigned arithmetic here (mod 1<<32).
-	 *
-	 * c = current_val = min_val	= the current value of the syncpoint.
-	 * t = thresh			= the value we are checking
-	 * f = future_val  = max_val	= the value c will reach when all
-	 *				  outstanding increments have completed.
-	 *
-	 * Note that c always chases f until it reaches f.
-	 *
-	 * Dtf = (f - t)
-	 * Dtc = (c - t)
-	 *
-	 *  Consider all cases:
-	 *
-	 *	A) .....c..t..f.....	Dtf < Dtc	need to wait
-	 *	B) .....c.....f..t..	Dtf > Dtc	expired
-	 *	C) ..t..c.....f.....	Dtf > Dtc	expired	   (Dct very large)
-	 *
-	 *  Any case where f==c: always expired (for any t).	Dtf == Dcf
-	 *  Any case where t==c: always expired (for any f).	Dtf >= Dtc (because Dtc==0)
-	 *  Any case where t==f!=c: always wait.		Dtf <  Dtc (because Dtf==0,
-	 *							Dtc!=0)
-	 *
-	 *  Other cases:
-	 *
-	 *	A) .....t..f..c.....	Dtf < Dtc	need to wait
-	 *	A) .....f..c..t.....	Dtf < Dtc	need to wait
-	 *	A) .....f..t..c.....	Dtf > Dtc	expired
-	 *
-	 *   So:
-	 *	   Dtf >= Dtc implies EXPIRED	(return true)
-	 *	   Dtf <  Dtc implies WAIT	(return false)
-	 *
-	 * Note: If t is expired then we *cannot* wait on it. We would wait
-	 * forever (hang the system).
-	 *
-	 * Note: do NOT get clever and remove the -thresh from both sides. It
-	 * is NOT the same.
-	 *
-	 * If future valueis zero, we have a client managed sync point. In that
-	 * case we do a direct comparison.
-	 */
-	if (!host1x_syncpt_client_managed(sp))
-		return future_val - thresh >= current_val - thresh;
-	else
-		return (s32)(current_val - thresh) >= 0;
+
+	return ((current_val - thresh) & 0x80000000U) == 0U;
 }
 
 int host1x_syncpt_init(struct host1x *host)
@@ -401,10 +370,15 @@ int host1x_syncpt_init(struct host1x *host)
 	host1x_hw_syncpt_enable_protection(host);
 
 	/* Allocate sync point to use for clearing waits for expired fences */
-	host->nop_sp = host1x_syncpt_alloc(host, NULL, 0);
+	host->nop_sp = host1x_syncpt_alloc(host, 0, "reserved-nop");
 	if (!host->nop_sp)
 		return -ENOMEM;
 
+	if (host->info->reserve_vblank_syncpts) {
+		kref_init(&host->syncpt[26].ref);
+		kref_init(&host->syncpt[27].ref);
+	}
+
 	return 0;
 }
 
@@ -416,44 +390,50 @@ int host1x_syncpt_init(struct host1x *host)
  * host1x client drivers can use this function to allocate a syncpoint for
  * subsequent use. A syncpoint returned by this function will be reserved for
  * use by the client exclusively. When no longer using a syncpoint, a host1x
- * client driver needs to release it using host1x_syncpt_free().
+ * client driver needs to release it using host1x_syncpt_put().
  */
 struct host1x_syncpt *host1x_syncpt_request(struct host1x_client *client,
 					    unsigned long flags)
 {
 	struct host1x *host = dev_get_drvdata(client->host->parent);
 
-	return host1x_syncpt_alloc(host, client, flags);
+	return host1x_syncpt_alloc(host, flags, dev_name(client->dev));
 }
 EXPORT_SYMBOL(host1x_syncpt_request);
 
-/**
- * host1x_syncpt_free() - free a requested syncpoint
- * @sp: host1x syncpoint
- *
- * Release a syncpoint previously allocated using host1x_syncpt_request(). A
- * host1x client driver should call this when the syncpoint is no longer in
- * use. Note that client drivers must ensure that the syncpoint doesn't remain
- * under the control of hardware after calling this function, otherwise two
- * clients may end up trying to access the same syncpoint concurrently.
- */
-void host1x_syncpt_free(struct host1x_syncpt *sp)
+static void syncpt_release(struct kref *ref)
 {
-	if (!sp)
-		return;
+	struct host1x_syncpt *sp = container_of(ref, struct host1x_syncpt, ref);
+
+	atomic_set(&sp->max_val, host1x_syncpt_read(sp));
 
 	mutex_lock(&sp->host->syncpt_mutex);
 
 	host1x_syncpt_base_free(sp->base);
 	kfree(sp->name);
 	sp->base = NULL;
-	sp->client = NULL;
 	sp->name = NULL;
 	sp->client_managed = false;
 
 	mutex_unlock(&sp->host->syncpt_mutex);
 }
-EXPORT_SYMBOL(host1x_syncpt_free);
+
+/**
+ * host1x_syncpt_put() - free a requested syncpoint
+ * @sp: host1x syncpoint
+ *
+ * Release a syncpoint previously allocated using host1x_syncpt_request(). A
+ * host1x client driver should call this when the syncpoint is no longer in
+ * use.
+ */
+void host1x_syncpt_put(struct host1x_syncpt *sp)
+{
+	if (!sp)
+		return;
+
+	kref_put(&sp->ref, syncpt_release);
+}
+EXPORT_SYMBOL(host1x_syncpt_put);
 
 void host1x_syncpt_deinit(struct host1x *host)
 {
@@ -520,16 +500,48 @@ unsigned int host1x_syncpt_nb_mlocks(struct host1x *host)
 }
 
 /**
- * host1x_syncpt_get() - obtain a syncpoint by ID
+ * host1x_syncpt_get_by_id() - obtain a syncpoint by ID
  * @host: host1x controller
  * @id: syncpoint ID
  */
-struct host1x_syncpt *host1x_syncpt_get(struct host1x *host, unsigned int id)
+struct host1x_syncpt *host1x_syncpt_get_by_id(struct host1x *host,
+					      unsigned int id)
 {
 	if (id >= host->info->nb_pts)
 		return NULL;
 
-	return host->syncpt + id;
+	if (kref_get_unless_zero(&host->syncpt[id].ref))
+		return &host->syncpt[id];
+	else
+		return NULL;
+}
+EXPORT_SYMBOL(host1x_syncpt_get_by_id);
+
+/**
+ * host1x_syncpt_get_by_id_noref() - obtain a syncpoint by ID but don't
+ * 	increase the refcount.
+ * @host: host1x controller
+ * @id: syncpoint ID
+ */
+struct host1x_syncpt *host1x_syncpt_get_by_id_noref(struct host1x *host,
+						    unsigned int id)
+{
+	if (id >= host->info->nb_pts)
+		return NULL;
+
+	return &host->syncpt[id];
+}
+EXPORT_SYMBOL(host1x_syncpt_get_by_id_noref);
+
+/**
+ * host1x_syncpt_get() - increment syncpoint refcount
+ * @sp: syncpoint
+ */
+struct host1x_syncpt *host1x_syncpt_get(struct host1x_syncpt *sp)
+{
+	kref_get(&sp->ref);
+
+	return sp;
 }
 EXPORT_SYMBOL(host1x_syncpt_get);
 
@@ -552,3 +564,31 @@ u32 host1x_syncpt_base_id(struct host1x_syncpt_base *base)
 	return base->id;
 }
 EXPORT_SYMBOL(host1x_syncpt_base_id);
+
+static void do_nothing(struct kref *ref)
+{
+}
+
+/**
+ * host1x_syncpt_release_vblank_reservation() - Make VBLANK syncpoint
+ *   available for allocation
+ *
+ * @client: host1x bus client
+ * @syncpt_id: syncpoint ID to make available
+ *
+ * Makes VBLANK<i> syncpoint available for allocatation if it was
+ * reserved at initialization time. This should be called by the display
+ * driver after it has ensured that any VBLANK increment programming configured
+ * by the boot chain has been disabled.
+ */
+void host1x_syncpt_release_vblank_reservation(struct host1x_client *client,
+					      u32 syncpt_id)
+{
+	struct host1x *host = dev_get_drvdata(client->host->parent);
+
+	if (!host->info->reserve_vblank_syncpts)
+		return;
+
+	kref_put(&host->syncpt[syncpt_id].ref, do_nothing);
+}
+EXPORT_SYMBOL(host1x_syncpt_release_vblank_reservation);
diff --git a/drivers/gpu/host1x/syncpt.h b/drivers/gpu/host1x/syncpt.h
index 8e1d04dacaa0..a6766f8d55ee 100644
--- a/drivers/gpu/host1x/syncpt.h
+++ b/drivers/gpu/host1x/syncpt.h
@@ -11,6 +11,7 @@
 #include <linux/atomic.h>
 #include <linux/host1x.h>
 #include <linux/kernel.h>
+#include <linux/kref.h>
 #include <linux/sched.h>
 
 #include "intr.h"
@@ -26,6 +27,8 @@ struct host1x_syncpt_base {
 };
 
 struct host1x_syncpt {
+	struct kref ref;
+
 	unsigned int id;
 	atomic_t min_val;
 	atomic_t max_val;
@@ -33,7 +36,6 @@ struct host1x_syncpt {
 	const char *name;
 	bool client_managed;
 	struct host1x *host;
-	struct host1x_client *client;
 	struct host1x_syncpt_base *base;
 
 	/* interrupt data */