gpu: nvgpu: move submit path to linux

Nvgpu submit path has a lot of dependency on Linux framework
e.g. use of copy_from_user, use of structures defined in uapi/nvgpu headers,
dma_buf_* calls for trace support etc

Hence to keep common code independent of Linux code, move submit path to
Linux directory

Move below APIs to common/linux/channel.c
trace_write_pushbuffer()
trace_write_pushbuffer_range()
gk20a_submit_prepare_syncs()
gk20a_submit_append_priv_cmdbuf()
gk20a_submit_append_gpfifo()
gk20a_submit_channel_gpfifo()

Move below APIs to common/linux/ce2.c
gk20a_ce_execute_ops()

Define gk20a_ce_execute_ops() in common/linux/ce2.c, and declare it in
gk20a/ce2_gk20a.h since it is needed in common/mm code too
Each OS needs to implement this API separately

gk20a_channel_alloc_gpfifo() use sizeof(nvgpu_gpfifo) to get size of one gpfifo
entry, but structure nvgpu_gpfifo is linux specific
Define new nvgpu_get_gpfifo_entry_size() in linux specific code and use it
in gk20a_channel_alloc_gpfifo() to get gpfifo entry size
Each OS needs to implement this API separately

Export some APIs from gk20a/ce2_gk20a.h and gk20a/channel_gk20a.h that are
needed in linux code

Jira NVGPU-259
Jira NVGPU-313

Change-Id: I360c6cb8ce4494b1e50c66af334a2a379f0d2dc4
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1586277
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

1 files changed, 1 insertions, 163 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/ce2_gk20a.c b/drivers/gpu/nvgpu/gk20a/ce2_gk20a.c
index 5314a1be..9ff6c792 100644
--- a/drivers/gpu/nvgpu/gk20a/ce2_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/ce2_gk20a.c
@@ -249,18 +249,7 @@ static inline unsigned int gk20a_ce_get_method_size(int request_operation,
         return methodsize;
 }
 
-static inline int gk20a_get_valid_launch_flags(struct gk20a *g, int launch_flags)
-{
-        /* there is no local memory available,
-        don't allow local memory related CE flags */
-        if (!g->mm.vidmem.size) {
-                launch_flags &= ~(NVGPU_CE_SRC_LOCATION_LOCAL_FB |
-                        NVGPU_CE_DST_LOCATION_LOCAL_FB);
-        }
-        return launch_flags;
-}
-
-static int gk20a_ce_prepare_submit(u64 src_buf,
+int gk20a_ce_prepare_submit(u64 src_buf,
                 u64 dst_buf,
                 u64 size,
                 u32 *cmd_buf_cpu_va,
@@ -626,157 +615,6 @@ end:
 }
 EXPORT_SYMBOL(gk20a_ce_create_context_with_cb);
 
-int gk20a_ce_execute_ops(struct gk20a *g,
-                u32 ce_ctx_id,
-                u64 src_buf,
-                u64 dst_buf,
-                u64 size,
-                unsigned int payload,
-                int launch_flags,
-                int request_operation,
-                struct gk20a_fence *gk20a_fence_in,
-                u32 submit_flags,
-                struct gk20a_fence **gk20a_fence_out)
-{
-        int ret = -EPERM;
-        struct gk20a_ce_app *ce_app = &g->ce_app;
-        struct gk20a_gpu_ctx *ce_ctx, *ce_ctx_save;
-        bool found = false;
-        u32 *cmd_buf_cpu_va;
-        u64 cmd_buf_gpu_va = 0;
-        u32 methodSize;
-        u32 cmd_buf_read_offset;
-        u32 fence_index;
-        struct nvgpu_gpfifo gpfifo;
-        struct nvgpu_fence fence = {0,0};
-        struct gk20a_fence *ce_cmd_buf_fence_out = NULL;
-        struct nvgpu_gpu_characteristics *gpu_capability = &g->gpu_characteristics;
-
-        if (!ce_app->initialised ||ce_app->app_state != NVGPU_CE_ACTIVE)
-                goto end;
-
-        nvgpu_mutex_acquire(&ce_app->app_mutex);
-
-        nvgpu_list_for_each_entry_safe(ce_ctx, ce_ctx_save,
-                        &ce_app->allocated_contexts, gk20a_gpu_ctx, list) {
-                if (ce_ctx->ctx_id == ce_ctx_id) {
-                        found = true;
-                        break;
-                }
-        }
-
-        nvgpu_mutex_release(&ce_app->app_mutex);
-
-        if (!found) {
-                ret = -EINVAL;
-                goto end;
-        }
-
-        if (ce_ctx->gpu_ctx_state != NVGPU_CE_GPU_CTX_ALLOCATED) {
-                ret = -ENODEV;
-                goto end;
-        }
-
-        nvgpu_mutex_acquire(&ce_ctx->gpu_ctx_mutex);
-
-        ce_ctx->cmd_buf_read_queue_offset %= ce_ctx->cmd_buf_end_queue_offset;
-
-        cmd_buf_read_offset = (ce_ctx->cmd_buf_read_queue_offset *
-                        (NVGPU_CE_MAX_COMMAND_BUFF_SIZE_PER_KICKOFF / sizeof(u32)));
-
-        /* at end of command buffer has gk20a_fence for command buffer sync */
-        fence_index = (cmd_buf_read_offset +
-                        ((NVGPU_CE_MAX_COMMAND_BUFF_SIZE_PER_KICKOFF / sizeof(u32)) -
-                        (NVGPU_CE_MAX_COMMAND_BUFF_SIZE_FOR_TRACING / sizeof(u32))));
-
-        if (sizeof(struct gk20a_fence *) > NVGPU_CE_MAX_COMMAND_BUFF_SIZE_FOR_TRACING) {
-                ret = -ENOMEM;
-                goto noop;
-        }
-
-        cmd_buf_cpu_va = (u32 *)ce_ctx->cmd_buf_mem.cpu_va;
-
-        /* 0 is treated as invalid pre-sync */
-        if (cmd_buf_cpu_va[fence_index]) {
-                struct gk20a_fence * ce_cmd_buf_fence_in = NULL;
-
-                memcpy((void *)&ce_cmd_buf_fence_in,
-                                (void *)(cmd_buf_cpu_va + fence_index),
-                                sizeof(struct gk20a_fence *));
-                ret = gk20a_fence_wait(g, ce_cmd_buf_fence_in,
-                                       gk20a_get_gr_idle_timeout(g));
-
-                gk20a_fence_put(ce_cmd_buf_fence_in);
-                /* Reset the stored last pre-sync */
-                memset((void *)(cmd_buf_cpu_va + fence_index),
-                                0,
-                                NVGPU_CE_MAX_COMMAND_BUFF_SIZE_FOR_TRACING);
-                if (ret)
-                        goto noop;
-        }
-
-        cmd_buf_gpu_va = (ce_ctx->cmd_buf_mem.gpu_va + (u64)(cmd_buf_read_offset *sizeof(u32)));
-
-        methodSize = gk20a_ce_prepare_submit(src_buf,
-                                        dst_buf,
-                                        size,
-                                        &cmd_buf_cpu_va[cmd_buf_read_offset],
-                                        NVGPU_CE_MAX_COMMAND_BUFF_SIZE_PER_KICKOFF,
-                                        payload,
-                                        gk20a_get_valid_launch_flags(g, launch_flags),
-                                        request_operation,
-                                        gpu_capability->dma_copy_class,
-                                        gk20a_fence_in);
-
-        if (methodSize) {
-                /* TODO: Remove CPU pre-fence wait */
-                if (gk20a_fence_in) {
-                        ret = gk20a_fence_wait(g, gk20a_fence_in,
-                                               gk20a_get_gr_idle_timeout(g));
-                        gk20a_fence_put(gk20a_fence_in);
-                        if (ret)
-                                goto noop;
-                }
-
-                /* store the element into gpfifo */
-                gpfifo.entry0 =
-                        u64_lo32(cmd_buf_gpu_va);
-                gpfifo.entry1 =
-                        (u64_hi32(cmd_buf_gpu_va) |
-                        pbdma_gp_entry1_length_f(methodSize));
-
-                /* take always the postfence as it is needed for protecting the ce context */
-                submit_flags |= NVGPU_SUBMIT_GPFIFO_FLAGS_FENCE_GET;
-
-                nvgpu_smp_wmb();
-
-                ret = gk20a_submit_channel_gpfifo(ce_ctx->ch, &gpfifo, NULL,
-                                        1, submit_flags, &fence,
-                                        &ce_cmd_buf_fence_out, false, NULL);
-
-                if (!ret) {
-                        memcpy((void *)(cmd_buf_cpu_va + fence_index),
-                                        (void *)&ce_cmd_buf_fence_out,
-                                        sizeof(struct gk20a_fence *));
-
-                        if (gk20a_fence_out) {
-                                gk20a_fence_get(ce_cmd_buf_fence_out);
-                                *gk20a_fence_out = ce_cmd_buf_fence_out;
-                        }
-
-                        /* Next available command buffer queue Index */
-                        ++ce_ctx->cmd_buf_read_queue_offset;
-                        ++ce_ctx->submitted_seq_number;
-                        }
-        } else
-                ret = -ENOMEM;
-noop:
-        nvgpu_mutex_release(&ce_ctx->gpu_ctx_mutex);
-end:
-        return ret;
-}
-EXPORT_SYMBOL(gk20a_ce_execute_ops);
-
 void gk20a_ce_delete_context(struct gk20a *g,
                 u32 ce_ctx_id)
 {