[PATCH 1/6] drm/amdgpu: Doorbell assignment for 8 sdma user queue per engine

[Felix Kuehling]

From: Shaoyun Liu <Shaoyun.Liu at amd.com>

Change doorbell assignments to allow routing doorbells for 8 user
mode SDMA queues per engine.

Signed-off-by: Shaoyun Liu <Shaoyun.Liu at amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling at amd.com>
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h             | 14 ++++-----
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c      | 38 +++++++++++++++----------
 drivers/gpu/drm/amd/include/kgd_kfd_interface.h |  6 ++--
 3 files changed, 33 insertions(+), 25 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index e992e0f..afa9e77 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -409,16 +409,16 @@ typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
 	AMDGPU_DOORBELL64_GFX_RING0               = 0x8b,
 
 	/*
-	 * Other graphics doorbells can be allocated here: from 0x8c to 0xef
+	 * Other graphics doorbells can be allocated here: from 0x8c to 0xdf
 	 * Graphics voltage island aperture 1
-	 * default non-graphics QWORD index is 0xF0 - 0xFF inclusive
+	 * default non-graphics QWORD index is 0xe0 - 0xFF inclusive
 	 */
 
-	/* sDMA engines */
-	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xF0,
-	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xF1,
-	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xF2,
-	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xF3,
+	/* sDMA engines  reserved from 0xe0 -oxef  */
+	AMDGPU_DOORBELL64_sDMA_ENGINE0            = 0xE0,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE0     = 0xE1,
+	AMDGPU_DOORBELL64_sDMA_ENGINE1            = 0xE8,
+	AMDGPU_DOORBELL64_sDMA_HI_PRI_ENGINE1     = 0xE9,
 
 	/* Interrupt handler */
 	AMDGPU_DOORBELL64_IH                      = 0xF4,  /* For legacy interrupt ring buffer */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index d7f08e3..bf0b012 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -126,7 +126,7 @@ static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
 
 void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 {
-	int i;
+	int i, n;
 	int last_valid_bit;
 	if (adev->kfd) {
 		struct kgd2kfd_shared_resources gpu_resources = {
@@ -165,7 +165,15 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 				&gpu_resources.doorbell_physical_address,
 				&gpu_resources.doorbell_aperture_size,
 				&gpu_resources.doorbell_start_offset);
-		if (adev->asic_type >= CHIP_VEGA10) {
+
+		if (adev->asic_type < CHIP_VEGA10) {
+			kgd2kfd->device_init(adev->kfd, &gpu_resources);
+			return;
+		}
+
+		n = (adev->asic_type < CHIP_VEGA20) ? 2 : 8;
+
+		for (i = 0; i < n; i += 2) {
 			/* On SOC15 the BIF is involved in routing
 			 * doorbells using the low 12 bits of the
 			 * address. Communicate the assignments to
@@ -173,20 +181,20 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 			 * process in case of 64-bit doorbells so we
 			 * can use each doorbell assignment twice.
 			 */
-			gpu_resources.sdma_doorbell[0][0] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE0;
-			gpu_resources.sdma_doorbell[0][1] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE0 + 0x200;
-			gpu_resources.sdma_doorbell[1][0] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE1;
-			gpu_resources.sdma_doorbell[1][1] =
-				AMDGPU_DOORBELL64_sDMA_ENGINE1 + 0x200;
-			/* Doorbells 0x0f0-0ff and 0x2f0-2ff are reserved for
-			 * SDMA, IH and VCN. So don't use them for the CP.
-			 */
-			gpu_resources.reserved_doorbell_mask = 0x1f0;
-			gpu_resources.reserved_doorbell_val  = 0x0f0;
+			gpu_resources.sdma_doorbell[0][i] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE0 + (i >> 1);
+			gpu_resources.sdma_doorbell[0][i+1] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE0 + 0x200 + (i >> 1);
+			gpu_resources.sdma_doorbell[1][i] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE1 + (i >> 1);
+			gpu_resources.sdma_doorbell[1][i+1] =
+				AMDGPU_DOORBELL64_sDMA_ENGINE1 + 0x200 + (i >> 1);
 		}
+		/* Doorbells 0x0e0-0ff and 0x2e0-2ff are reserved for
+		 * SDMA, IH and VCN. So don't use them for the CP.
+		 */
+		gpu_resources.reserved_doorbell_mask = 0x1e0;
+		gpu_resources.reserved_doorbell_val  = 0x0e0;
 
 		kgd2kfd->device_init(adev->kfd, &gpu_resources);
 	}
diff --git a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
index 4ef0a50..0545888 100644
--- a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
+++ b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
@@ -146,10 +146,10 @@ struct kgd2kfd_shared_resources {
 	 * is reserved: (D & reserved_doorbell_mask) == reserved_doorbell_val
 	 *
 	 * KFD currently uses 1024 (= 0x3ff) doorbells per process. If
-	 * doorbells 0x0f0-0x0f7 and 0x2f-0x2f7 are reserved, that means
-	 * mask would be set to 0x1f8 and val set to 0x0f0.
+	 * doorbells 0x0e0-0x0ff and 0x2e0-0x2ff are reserved, that means
+	 * mask would be set to 0x1e0 and val set to 0x0e0.
 	 */
-	unsigned int sdma_doorbell[2][2];
+	unsigned int sdma_doorbell[2][8];
 	unsigned int reserved_doorbell_mask;
 	unsigned int reserved_doorbell_val;
 
-- 
2.7.4