[PATCH v1] drm/amdgpu: optimize amdgpu driver load & resume time

[S, Shirish]

From: Shirish S <shirish.s at amd.com>

amdgpu_device_resume() & amdgpu_device_init() have a high
time consuming call of amdgpu_late_init() which sets the
clock_gating state of all IP blocks and is blocking.
This patch defers only this setting of clock gating state
operation to post resume of amdgpu driver but ideally before
the UI comes up or in some cases post ui as well.

With this change the resume time of amdgpu_device comes down
from 1.299s to 0.199s which further helps in reducing the overall
system resume time.

V1: made the optimization applicable during driver load as well.

TEST:(For ChromiumOS on STONEY only)
* UI comes up
* S3 works multiple times
* Resume time is consistent and lower
* amdgpu_late_init() call gets called consistently and no errors reported.

Signed-off-by: Shirish S <shirish.s at amd.com>
Reviewed-by: Huang Rui <ray.huang at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  3 ++
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 46 +++++++++++++++++++++++-------
 2 files changed, 39 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 7a20a0b..2474332 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1616,6 +1616,9 @@ struct amdgpu_device {
 	/* amdkfd interface */
 	struct kfd_dev          *kfd;
 
+	/* delayed work_func for deferring clockgating during resume */
+	struct delayed_work     late_init_work;
+
 	struct amdgpu_virt	virt;
 
 	/* link all shadow bo */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index cfc650c..224d3eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -57,6 +57,8 @@
 
 MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
 
+#define AMDGPU_RESUME_MS		2000
+
 static int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
 static void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
 
@@ -1655,22 +1657,13 @@ static int amdgpu_init(struct amdgpu_device *adev)
 	return 0;
 }
 
-static int amdgpu_late_init(struct amdgpu_device *adev)
+static int amdgpu_late_set_cg_state(struct amdgpu_device *adev)
 {
 	int i = 0, r;
 
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
-		if (adev->ip_blocks[i].version->funcs->late_init) {
-			r = adev->ip_blocks[i].version->funcs->late_init((void *)adev);
-			if (r) {
-				DRM_ERROR("late_init of IP block <%s> failed %d\n",
-					  adev->ip_blocks[i].version->funcs->name, r);
-				return r;
-			}
-			adev->ip_blocks[i].status.late_initialized = true;
-		}
 		/* skip CG for VCE/UVD, it's handled specially */
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
 		    adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE) {
@@ -1688,6 +1681,29 @@ static int amdgpu_late_init(struct amdgpu_device *adev)
 	return 0;
 }
 
+static int amdgpu_late_init(struct amdgpu_device *adev)
+{
+	int i = 0, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid)
+			continue;
+		if (adev->ip_blocks[i].version->funcs->late_init) {
+			r = adev->ip_blocks[i].version->funcs->late_init((void *)adev);
+			if (r) {
+				DRM_ERROR("late_init of IP block <%s> failed %d\n",
+					  adev->ip_blocks[i].version->funcs->name, r);
+				return r;
+			}
+			adev->ip_blocks[i].status.late_initialized = true;
+		}
+	}
+
+	mod_delayed_work(system_wq, &adev->late_init_work,
+			msecs_to_jiffies(AMDGPU_RESUME_MS));
+	return 0;
+}
+
 static int amdgpu_fini(struct amdgpu_device *adev)
 {
 	int i, r;
@@ -1775,6 +1791,13 @@ static int amdgpu_fini(struct amdgpu_device *adev)
 	return 0;
 }
 
+static void amdgpu_late_init_func_handler(struct work_struct *work)
+{
+	struct amdgpu_device *adev =
+		container_of(work, struct amdgpu_device, late_init_work.work);
+	amdgpu_late_set_cg_state(adev);
+}
+
 int amdgpu_suspend(struct amdgpu_device *adev)
 {
 	int i, r;
@@ -2074,6 +2097,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	INIT_LIST_HEAD(&adev->gtt_list);
 	spin_lock_init(&adev->gtt_list_lock);
 
+	INIT_DELAYED_WORK(&adev->late_init_work, amdgpu_late_init_func_handler);
+
 	if (adev->asic_type >= CHIP_BONAIRE) {
 		adev->rmmio_base = pci_resource_start(adev->pdev, 5);
 		adev->rmmio_size = pci_resource_len(adev->pdev, 5);
@@ -2270,6 +2295,7 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
 	amdgpu_fbdev_fini(adev);
 	r = amdgpu_fini(adev);
 	adev->accel_working = false;
+	cancel_delayed_work_sync(&adev->late_init_work);
 	/* free i2c buses */
 	if (!amdgpu_device_has_dc_support(adev))
 		amdgpu_i2c_fini(adev);
-- 
2.7.4