nv-tegra-mirror/linux-nv-oot
1
0
Fork 0
mirror of git://nv-tegra.nvidia.com/linux-nv-oot.git synced 2025-12-24 10:11:26 +03:00
Code Issues Packages Projects Releases Wiki Activity

video: tegra: host: nvdla: add offset to memory handles

Browse Source 
Use offset and memory handle to calculate IOVA address
for descriptors. It allows using single buffer in user
space for all descriptors.

Jira DLA-176

Change-Id: I141efa7fc8662be8aa4b5c3bd2ea7a369a90769a
Signed-off-by: Prashant Gaikwad <pgaikwad@nvidia.com>
Reviewed-on: http://git-master/r/1220956
This commit is contained in:
Prashant Gaikwad
2016-09-15 01:29:18 +05:30
committed by Laxman Dewangan
parent cbadc2aa0c
commit fd0023d8df
2 changed files with 39 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
drivers/video/tegra/host/nvdla/nvdla.h
View File

@@ -33,7 +33,6 @@
#define NV_DLA_THI_METHOD_DATA 0x00000044 /* RW-4R */
#define MAX_NUM_ACTION_LIST 1
#define MAX_HANDLE_PER_OP_DESC 1
/* TODO: Below should come from firmware interface */
#define ACTION_OPCODE_TERMINATE 0x00
@@ -113,7 +112,8 @@ struct nvdla_task {
dma_addr_t task_desc_pa;
size_t buf_size;
int timeout;
u32 *op_handle;
u32 *memory_handles;
u32 num_handles;
};
extern const struct file_operations tegra_nvdla_ctrl_ops;

55
drivers/video/tegra/host/nvdla/nvdla_queue.c
View File

@@ -121,7 +121,7 @@ static void task_free(struct kref *ref)
}
/* free operation descriptor handle */
kfree(task->op_handle);
kfree(task->memory_handles);
/* finally free task */
kfree(task);
@@ -151,7 +151,7 @@ static void nvdla_task_free_locked(struct nvdla_task *task)
/* unpin submit ref */
nvhost_buffer_submit_unpin(task->buffers,
task->op_handle, MAX_HANDLE_PER_OP_DESC);
task->memory_handles, task->num_handles);
/* update takslist */
list_del(&task->list);
@@ -235,11 +235,11 @@ struct nvdla_task *nvdla_task_alloc(struct nvhost_queue *queue,
size_t preactionlist_size;
uint16_t postactionl_of;
uint16_t preactionl_of;
dma_addr_t op_dma_addr;
dma_addr_t *dma_addr;
dma_addr_t buffer_pa;
size_t task_size;
size_t buf_size;
size_t op_size;
size_t *dma_size;
u32 *buffer_va;
void *mem;
int err;
@@ -429,35 +429,54 @@ struct nvdla_task *nvdla_task_alloc(struct nvhost_queue *queue,
if (num_operations) {
task->buffers = buffers;
task->op_handle = kcalloc(MAX_HANDLE_PER_OP_DESC, sizeof(u32),
task->num_handles = 2;
task->memory_handles = kcalloc(task->num_handles, sizeof(u32),
GFP_KERNEL);
if (!task->op_handle) {
if (!task->memory_handles) {
err = -ENOMEM;
goto fail_to_alloc_opdesc;
goto fail_to_alloc_handles;
}
if (copy_from_user(task->op_handle,
(void __user *)user_task.operation_desc,
(MAX_HANDLE_PER_OP_DESC * sizeof(u32)))) {
err = -EFAULT;
goto fail_to_cpy_buffer;
dma_addr = kcalloc(task->num_handles, sizeof(dma_addr_t),
GFP_KERNEL);
if (!dma_addr) {
err = -ENOMEM;
goto fail_to_alloc_dma_addr;
}
err = nvhost_buffer_submit_pin(buffers, task->op_handle,
MAX_HANDLE_PER_OP_DESC, &op_dma_addr, &op_size);
dma_size = kcalloc(task->num_handles, sizeof(u32),
GFP_KERNEL);
if (!dma_size) {
err = -ENOMEM;
goto fail_to_alloc_dma_size;
}
task->memory_handles[0] = user_task.operation_desc.handle;
task->memory_handles[1] = user_task.surface_desc.handle;
err = nvhost_buffer_submit_pin(buffers, task->memory_handles,
task->num_handles, dma_addr, dma_size);
if (!err) {
task_desc->operation_desc = op_dma_addr;
task_desc->operation_desc = dma_addr[0] +
user_task.operation_desc.offset;
task_desc->surface_desc = dma_addr[1] +
user_task.surface_desc.offset;
task_desc->num_operations = num_operations;
}
kfree(dma_addr);
kfree(dma_size);
}
nvdla_dbg_info(pdev, "task[%p] initialized", task);
return task;
fail_to_cpy_buffer:
kfree(task->op_handle);
fail_to_alloc_opdesc:
fail_to_alloc_dma_size:
kfree(dma_addr);
fail_to_alloc_dma_addr:
kfree(task->memory_handles);
fail_to_alloc_handles:
fail_to_dma_alloc:
kfree(task);
fail_to_alloc_task: