nv-tegra-mirror/linux-nvgpu
1
0
Fork 0
mirror of git://nv-tegra.nvidia.com/linux-nvgpu.git synced 2025-12-25 02:52:51 +03:00
Code Issues Packages Projects Releases Wiki Activity

gpu: nvgpu: add gr/ctx and gr/subctx APIs to set hwpm ctxsw mode

Browse Source 
gr_gk20a_update_hwpm_ctxsw_mode() right now validates the incoming
hwpm mode, checks if it is already set, and if not, it will go ahead
and set the new hwpm mode by calling g->ops.gr.ctxsw_prog HALs

Instead of programming hwpm mode in gr_gk20a.c, move the programming
to gr/ctx and gr/subctx units by adding below APIs
nvgpu_gr_ctx_prepare_hwpm_mode() - validate the incoming mode and
                                   check if it is already set
nvgpu_gr_ctx_set_hwpm_mode() - set pm mode in graphics context
nvgpu_gr_subctx_set_hwpm_mode() - set pm mode in subcontext

Add gpu_va field to struct pm_ctx_desc to store the gpu_va to be
programmed into context

Rename NVGPU_DBG_HWPM_CTXSW_MODE_* to NVGPU_GR_CTX_HWPM_CTXSW_MODE_*
and move them to gr/ctx.h

Remove below HALs since they are no longer used
g->ops.gr.ctxsw_prog.set_pm_mode_no_ctxsw()
g->ops.gr.ctxsw_prog.set_pm_mode_ctxsw()
g->ops.gr.ctxsw_prog.set_pm_mode_stream_out_ctxsw()

Jira NVGPU-1527
Jira NVGPU-1613

Change-Id: Id2a4d498182ec0e3586dc7265f73a25870ca2ef7
Signed-off-by: Deepak Nibade <dnibade@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2011093
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This commit is contained in:
Deepak Nibade
2019-01-24 15:01:52 +05:30
committed by mobile promotions
parent dd12b9b320
commit fe27a7f934
20 changed files with 147 additions and 173 deletions
Download Patch File Download Diff File Expand all files Collapse all files

80
drivers/gpu/nvgpu/common/gr/ctx.c
View File

@@ -673,3 +673,83 @@ int nvgpu_gr_ctx_set_smpc_mode(struct gk20a *g, struct nvgpu_gr_ctx *gr_ctx,
return err;
}
int nvgpu_gr_ctx_prepare_hwpm_mode(struct gk20a *g, struct nvgpu_gr_ctx *gr_ctx,
u32 mode, bool *skip_update)
{
struct pm_ctx_desc *pm_ctx = &gr_ctx->pm_ctx;
*skip_update = false;
if (!nvgpu_mem_is_valid(&gr_ctx->mem)) {
nvgpu_err(g, "no graphics context allocated");
return -EFAULT;
}
if ((mode == NVGPU_GR_CTX_HWPM_CTXSW_MODE_STREAM_OUT_CTXSW) &&
(g->ops.gr.ctxsw_prog.hw_get_pm_mode_stream_out_ctxsw == NULL)) {
nvgpu_err(g,
"Mode-E hwpm context switch mode is not supported");
return -EINVAL;
}
switch (mode) {
case NVGPU_GR_CTX_HWPM_CTXSW_MODE_CTXSW:
if (pm_ctx->pm_mode ==
g->ops.gr.ctxsw_prog.hw_get_pm_mode_ctxsw()) {
*skip_update = true;
return 0;
}
pm_ctx->pm_mode = g->ops.gr.ctxsw_prog.hw_get_pm_mode_ctxsw();
pm_ctx->gpu_va = pm_ctx->mem.gpu_va;
break;
case NVGPU_GR_CTX_HWPM_CTXSW_MODE_NO_CTXSW:
if (pm_ctx->pm_mode ==
g->ops.gr.ctxsw_prog.hw_get_pm_mode_no_ctxsw()) {
*skip_update = true;
return 0;
}
pm_ctx->pm_mode =
g->ops.gr.ctxsw_prog.hw_get_pm_mode_no_ctxsw();
pm_ctx->gpu_va = 0;
break;
case NVGPU_GR_CTX_HWPM_CTXSW_MODE_STREAM_OUT_CTXSW:
if (pm_ctx->pm_mode ==
g->ops.gr.ctxsw_prog.hw_get_pm_mode_stream_out_ctxsw()) {
*skip_update = true;
return 0;
}
pm_ctx->pm_mode =
g->ops.gr.ctxsw_prog.hw_get_pm_mode_stream_out_ctxsw();
pm_ctx->gpu_va = pm_ctx->mem.gpu_va;
break;
default:
nvgpu_err(g, "invalid hwpm context switch mode");
return -EINVAL;
}
return 0;
}
int nvgpu_gr_ctx_set_hwpm_mode(struct gk20a *g, struct nvgpu_gr_ctx *gr_ctx,
bool set_pm_ptr)
{
int err;
/* Channel gr_ctx buffer is gpu cacheable.
Flush and invalidate before cpu update. */
err = g->ops.mm.l2_flush(g, true);
if (err != 0) {
nvgpu_err(g, "l2_flush failed");
return err;
}
g->ops.gr.ctxsw_prog.set_pm_mode(g, &gr_ctx->mem,
gr_ctx->pm_ctx.pm_mode);
if (set_pm_ptr) {
g->ops.gr.ctxsw_prog.set_pm_ptr(g, &gr_ctx->mem,
gr_ctx->pm_ctx.gpu_va);
}
return err;
}

17
drivers/gpu/nvgpu/common/gr/ctxsw_prog/ctxsw_prog_gm20b.c
View File

@@ -141,23 +141,6 @@ void gm20b_ctxsw_prog_set_pm_smpc_mode(struct gk20a *g,
nvgpu_mem_wr(g, ctx_mem, ctxsw_prog_main_image_pm_o(), data);
}
u32 gm20b_ctxsw_prog_set_pm_mode_no_ctxsw(struct gk20a *g,
struct nvgpu_mem *ctx_mem)
{
gm20b_ctxsw_prog_set_pm_mode(g, ctx_mem,
ctxsw_prog_main_image_pm_mode_no_ctxsw_f());
return ctxsw_prog_main_image_pm_mode_no_ctxsw_f();
}
u32 gm20b_ctxsw_prog_set_pm_mode_ctxsw(struct gk20a *g,
struct nvgpu_mem *ctx_mem)
{
gm20b_ctxsw_prog_set_pm_mode(g, ctx_mem,
ctxsw_prog_main_image_pm_mode_ctxsw_f());
return ctxsw_prog_main_image_pm_mode_ctxsw_f();
}
u32 gm20b_ctxsw_prog_hw_get_pm_mode_no_ctxsw(void)
{
return ctxsw_prog_main_image_pm_mode_no_ctxsw_f();

4
drivers/gpu/nvgpu/common/gr/ctxsw_prog/ctxsw_prog_gm20b.h
View File

@@ -52,10 +52,6 @@ void gm20b_ctxsw_prog_set_pm_mode(struct gk20a *g,
struct nvgpu_mem *ctx_mem, u32 mode);
void gm20b_ctxsw_prog_set_pm_smpc_mode(struct gk20a *g,
struct nvgpu_mem *ctx_mem, bool enable);
u32 gm20b_ctxsw_prog_set_pm_mode_no_ctxsw(struct gk20a *g,
struct nvgpu_mem *ctx_mem);
u32 gm20b_ctxsw_prog_set_pm_mode_ctxsw(struct gk20a *g,
struct nvgpu_mem *ctx_mem);
u32 gm20b_ctxsw_prog_hw_get_pm_mode_no_ctxsw(void);
u32 gm20b_ctxsw_prog_hw_get_pm_mode_ctxsw(void);
void gm20b_ctxsw_prog_init_ctxsw_hdr_data(struct gk20a *g,

8
drivers/gpu/nvgpu/common/gr/ctxsw_prog/ctxsw_prog_gv11b.c
View File

@@ -54,14 +54,6 @@ u32 gv11b_ctxsw_prog_hw_get_pm_mode_stream_out_ctxsw(void)
return ctxsw_prog_main_image_pm_mode_stream_out_ctxsw_f();
}
u32 gv11b_ctxsw_prog_set_pm_mode_stream_out_ctxsw(struct gk20a *g,
struct nvgpu_mem *ctx_mem)
{
gm20b_ctxsw_prog_set_pm_mode(g, ctx_mem,
ctxsw_prog_main_image_pm_mode_stream_out_ctxsw_f());
return ctxsw_prog_main_image_pm_mode_stream_out_ctxsw_f();
}
void gv11b_ctxsw_prog_set_full_preemption_ptr(struct gk20a *g,
struct nvgpu_mem *ctx_mem, u64 addr)
{

2
drivers/gpu/nvgpu/common/gr/ctxsw_prog/ctxsw_prog_gv11b.h
View File

@@ -30,8 +30,6 @@ void gv11b_ctxsw_prog_set_zcull_ptr(struct gk20a *g, struct nvgpu_mem *ctx_mem,
void gv11b_ctxsw_prog_set_pm_ptr(struct gk20a *g, struct nvgpu_mem *ctx_mem,
u64 addr);
u32 gv11b_ctxsw_prog_hw_get_pm_mode_stream_out_ctxsw(void);
u32 gv11b_ctxsw_prog_set_pm_mode_stream_out_ctxsw(struct gk20a *g,
struct nvgpu_mem *ctx_mem);
void gv11b_ctxsw_prog_set_full_preemption_ptr(struct gk20a *g,
struct nvgpu_mem *ctx_mem, u64 addr);
void gv11b_ctxsw_prog_set_full_preemption_ptr_veid0(struct gk20a *g,

7
drivers/gpu/nvgpu/common/gr/subctx.c
View File

@@ -117,3 +117,10 @@ void nvgpu_gr_subctx_zcull_setup(struct gk20a *g, struct nvgpu_gr_subctx *subctx
g->ops.gr.ctxsw_prog.set_zcull_ptr(g, &subctx->ctx_header,
gr_ctx->zcull_ctx.gpu_va);
}
void nvgpu_gr_subctx_set_hwpm_mode(struct gk20a *g,
struct nvgpu_gr_subctx *subctx, struct nvgpu_gr_ctx *gr_ctx)
{
g->ops.gr.ctxsw_prog.set_pm_ptr(g, &subctx->ctx_header,
gr_ctx->pm_ctx.gpu_va);
}