[PATCH 2/2] drm/amdkfd: Expose HDP registers to user space

[Zeng, Oak]

Introduce a new memory type (KFD_IOC_ALLOC_MEM_FLAGS_HDP_FLUSH) and
expose mmio page of HDP registers to user space through this new
memory type.

v2: moved remapped hdp regs to adev struct

Change-Id: If5ac13c46ea7fbd2194ddc8b2ece26ef4f76c330
Signed-off-by: Oak Zeng <Oak.Zeng at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c       | 7 +++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h       | 1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 7 ++++---
 drivers/gpu/drm/amd/amdkfd/kfd_chardev.c         | 4 ++++
 drivers/gpu/drm/amd/include/kgd_kfd_interface.h  | 1 +
 include/uapi/linux/kfd_ioctl.h                   | 1 +
 6 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index acf8ae0..d953338b8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -482,6 +482,13 @@ uint64_t amdgpu_amdkfd_get_hive_id(struct kgd_dev *kgd)
 	return adev->gmc.xgmi.hive_id;
 }
 
+uint64_t amdgpu_amdkfd_get_hdp_register_physical_addr(struct kgd_dev *kgd)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+
+	return adev->remapped_hdp_reg.remapped_hdp_mem_flush_cntl_physical_addr;
+}
+
 int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
 				uint32_t vmid, uint64_t gpu_addr,
 				uint32_t *ib_cmd, uint32_t ib_len)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index e6a5037..e778679 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -154,6 +154,7 @@ int amdgpu_amdkfd_get_dmabuf_info(struct kgd_dev *kgd, int dma_buf_fd,
 				  uint32_t *flags);
 uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd);
 uint64_t amdgpu_amdkfd_get_hive_id(struct kgd_dev *kgd);
+uint64_t amdgpu_amdkfd_get_hdp_register_physical_addr(struct kgd_dev *kgd);
 
 #define read_user_wptr(mmptr, wptr, dst)				\
 	({								\
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 047bba8..4394f61 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -1087,7 +1087,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
 		if (!offset || !*offset)
 			return -EINVAL;
 		user_addr = *offset;
-	} else if (flags & ALLOC_MEM_FLAGS_DOORBELL) {
+	} else if (flags & (ALLOC_MEM_FLAGS_DOORBELL |
+			ALLOC_MEM_FLAGS_HDP_FLUSH)) {
 		domain = AMDGPU_GEM_DOMAIN_GTT;
 		alloc_domain = AMDGPU_GEM_DOMAIN_CPU;
 		bo_type = ttm_bo_type_sg;
@@ -1263,8 +1264,8 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 	/* Free the sync object */
 	amdgpu_sync_free(&mem->sync);
 
-	/* If the SG is not NULL, it's one we created for a doorbell
-	 * BO. We need to free it.
+	/* If the SG is not NULL, it's one we created for a doorbell or hdp
+	 * flush BO. We need to free it.
 	 */
 	if (mem->bo->tbo.sg) {
 		sg_free_table(mem->bo->tbo.sg);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 083bd81..973d2fe 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -1272,6 +1272,10 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep,
 		if (args->size != kfd_doorbell_process_slice(dev))
 			return -EINVAL;
 		offset = kfd_get_process_doorbells(dev, p);
+	} else if (flags & KFD_IOC_ALLOC_MEM_FLAGS_HDP_FLUSH) {
+		if (args->size != PAGE_SIZE)
+			return -EINVAL;
+		offset = amdgpu_amdkfd_get_hdp_register_physical_addr(dev->kgd);
 	}
 
 	mutex_lock(&p->mutex);
diff --git a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
index 5f3c10e..3bc78b26 100644
--- a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
+++ b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
@@ -186,6 +186,7 @@ struct tile_config {
 #define ALLOC_MEM_FLAGS_GTT		(1 << 1)
 #define ALLOC_MEM_FLAGS_USERPTR		(1 << 2)
 #define ALLOC_MEM_FLAGS_DOORBELL	(1 << 3)
+#define ALLOC_MEM_FLAGS_HDP_FLUSH	(1 << 4)
 
 /*
  * Allocation flags attributes/access options.
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h
index dc067ed..96afd01 100644
--- a/include/uapi/linux/kfd_ioctl.h
+++ b/include/uapi/linux/kfd_ioctl.h
@@ -338,6 +338,7 @@ struct kfd_ioctl_acquire_vm_args {
 #define KFD_IOC_ALLOC_MEM_FLAGS_GTT		(1 << 1)
 #define KFD_IOC_ALLOC_MEM_FLAGS_USERPTR		(1 << 2)
 #define KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL	(1 << 3)
+#define KFD_IOC_ALLOC_MEM_FLAGS_HDP_FLUSH	(1 << 4)
 /* Allocation flags: attributes/access options */
 #define KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE	(1 << 31)
 #define KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE	(1 << 30)
-- 
2.7.4