[PATCH] drm/amdgpu/display: fix FP handling in DCN30

[PATCH] drm/amdgpu/display: fix FP handling in DCN30

[Bhawanpreet Lakha]

From: Alex Deucher <alexander.deucher@amd.com>

Adjust the FP handling to avoid nested calls.

The nested calls cause the warning below
WARNING: CPU: 3 PID: 384 at arch/x86/kernel/fpu/core.c:129 kernel_fpu_begin

Fixes: 26803606c5d6 ("drm/amdgpu/display: FP fixes for DCN3.x (v4)")
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
---
 .../drm/amd/display/dc/dcn30/dcn30_resource.c | 43 +++----------------
 1 file changed, 6 insertions(+), 37 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
index b379057e669c..d5c81ad55045 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
@@ -1470,20 +1470,8 @@ int dcn30_populate_dml_pipes_from_context(
 	return pipe_cnt;
 }
 
-/*
- * This must be noinline to ensure anything that deals with FP registers
- * is contained within this call; previously our compiling with hard-float
- * would result in fp instructions being emitted outside of the boundaries
- * of the DC_FP_START/END macros, which makes sense as the compiler has no
- * idea about what is wrapped and what is not
- *
- * This is largely just a workaround to avoid breakage introduced with 5.6,
- * ideally all fp-using code should be moved into its own file, only that
- * should be compiled with hard-float, and all code exported from there
- * should be strictly wrapped with DC_FP_START/END
- */
-static noinline void dcn30_populate_dml_writeback_from_context_fp(
-		struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
+void dcn30_populate_dml_writeback_from_context(
+	struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
 {
 	int pipe_cnt, i, j;
 	double max_calc_writeback_dispclk;
@@ -1571,14 +1559,6 @@ static noinline void dcn30_populate_dml_writeback_from_context_fp(
 
 }
 
-void dcn30_populate_dml_writeback_from_context(
-		struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
-{
-	DC_FP_START();
-	dcn30_populate_dml_writeback_from_context_fp(dc, res_ctx, pipes);
-	DC_FP_END();
-}
-
 unsigned int dcn30_calc_max_scaled_time(
 		unsigned int time_per_pixel,
 		enum mmhubbub_wbif_mode mode,
@@ -1977,7 +1957,7 @@ static struct pipe_ctx *dcn30_find_split_pipe(
 	return pipe;
 }
 
-static bool dcn30_internal_validate_bw(
+static noinline bool dcn30_internal_validate_bw(
 		struct dc *dc,
 		struct dc_state *context,
 		display_e2e_pipe_params_st *pipes,
@@ -1999,6 +1979,7 @@ static bool dcn30_internal_validate_bw(
 
 	pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate);
 
+	DC_FP_START();
 	if (!pipe_cnt) {
 		out = true;
 		goto validate_out;
@@ -2222,6 +2203,7 @@ static bool dcn30_internal_validate_bw(
 	out = false;
 
 validate_out:
+	DC_FP_END();
 	return out;
 }
 
@@ -2404,7 +2386,7 @@ void dcn30_calculate_wm_and_dlg(
 	DC_FP_END();
 }
 
-static noinline bool dcn30_validate_bandwidth_fp(struct dc *dc,
+bool dcn30_validate_bandwidth(struct dc *dc,
 		struct dc_state *context,
 		bool fast_validate)
 {
@@ -2455,19 +2437,6 @@ static noinline bool dcn30_validate_bandwidth_fp(struct dc *dc,
 	return out;
 }
 
-bool dcn30_validate_bandwidth(struct dc *dc,
-		struct dc_state *context,
-		bool fast_validate)
-{
-	bool out;
-
-	DC_FP_START();
-	out = dcn30_validate_bandwidth_fp(dc, context, fast_validate);
-	DC_FP_END();
-
-	return out;
-}
-
 static noinline void get_optimal_dcfclk_fclk_for_uclk(unsigned int uclk_mts,
                                                        unsigned int *optimal_dcfclk,
                                                        unsigned int *optimal_fclk)
-- 
2.25.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH] drm/amdgpu/display: fix FP handling in DCN30

[Alex Deucher]

On Thu, Nov 12, 2020 at 5:06 PM Bhawanpreet Lakha
<Bhawanpreet.Lakha@amd.com> wrote:
>
> From: Alex Deucher <alexander.deucher@amd.com>
>
> Adjust the FP handling to avoid nested calls.
>
> The nested calls cause the warning below
> WARNING: CPU: 3 PID: 384 at arch/x86/kernel/fpu/core.c:129 kernel_fpu_begin
>
> Fixes: 26803606c5d6 ("drm/amdgpu/display: FP fixes for DCN3.x (v4)")
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>

Looks good to me.  Thanks!

Alex

> ---
>  .../drm/amd/display/dc/dcn30/dcn30_resource.c | 43 +++----------------
>  1 file changed, 6 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
> index b379057e669c..d5c81ad55045 100644
> --- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
> +++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
> @@ -1470,20 +1470,8 @@ int dcn30_populate_dml_pipes_from_context(
>         return pipe_cnt;
>  }
>
> -/*
> - * This must be noinline to ensure anything that deals with FP registers
> - * is contained within this call; previously our compiling with hard-float
> - * would result in fp instructions being emitted outside of the boundaries
> - * of the DC_FP_START/END macros, which makes sense as the compiler has no
> - * idea about what is wrapped and what is not
> - *
> - * This is largely just a workaround to avoid breakage introduced with 5.6,
> - * ideally all fp-using code should be moved into its own file, only that
> - * should be compiled with hard-float, and all code exported from there
> - * should be strictly wrapped with DC_FP_START/END
> - */
> -static noinline void dcn30_populate_dml_writeback_from_context_fp(
> -               struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
> +void dcn30_populate_dml_writeback_from_context(
> +       struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
>  {
>         int pipe_cnt, i, j;
>         double max_calc_writeback_dispclk;
> @@ -1571,14 +1559,6 @@ static noinline void dcn30_populate_dml_writeback_from_context_fp(
>
>  }
>
> -void dcn30_populate_dml_writeback_from_context(
> -               struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
> -{
> -       DC_FP_START();
> -       dcn30_populate_dml_writeback_from_context_fp(dc, res_ctx, pipes);
> -       DC_FP_END();
> -}
> -
>  unsigned int dcn30_calc_max_scaled_time(
>                 unsigned int time_per_pixel,
>                 enum mmhubbub_wbif_mode mode,
> @@ -1977,7 +1957,7 @@ static struct pipe_ctx *dcn30_find_split_pipe(
>         return pipe;
>  }
>
> -static bool dcn30_internal_validate_bw(
> +static noinline bool dcn30_internal_validate_bw(
>                 struct dc *dc,
>                 struct dc_state *context,
>                 display_e2e_pipe_params_st *pipes,
> @@ -1999,6 +1979,7 @@ static bool dcn30_internal_validate_bw(
>
>         pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate);
>
> +       DC_FP_START();
>         if (!pipe_cnt) {
>                 out = true;
>                 goto validate_out;
> @@ -2222,6 +2203,7 @@ static bool dcn30_internal_validate_bw(
>         out = false;
>
>  validate_out:
> +       DC_FP_END();
>         return out;
>  }
>
> @@ -2404,7 +2386,7 @@ void dcn30_calculate_wm_and_dlg(
>         DC_FP_END();
>  }
>
> -static noinline bool dcn30_validate_bandwidth_fp(struct dc *dc,
> +bool dcn30_validate_bandwidth(struct dc *dc,
>                 struct dc_state *context,
>                 bool fast_validate)
>  {
> @@ -2455,19 +2437,6 @@ static noinline bool dcn30_validate_bandwidth_fp(struct dc *dc,
>         return out;
>  }
>
> -bool dcn30_validate_bandwidth(struct dc *dc,
> -               struct dc_state *context,
> -               bool fast_validate)
> -{
> -       bool out;
> -
> -       DC_FP_START();
> -       out = dcn30_validate_bandwidth_fp(dc, context, fast_validate);
> -       DC_FP_END();
> -
> -       return out;
> -}
> -
>  static noinline void get_optimal_dcfclk_fclk_for_uclk(unsigned int uclk_mts,
>                                                         unsigned int *optimal_dcfclk,
>                                                         unsigned int *optimal_fclk)
> --
> 2.25.1
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH] drm/amdgpu/display: fix FP handling in DCN30

[Kazlauskas, Nicholas]

On 2020-11-12 5:06 p.m., Bhawanpreet Lakha wrote:
> From: Alex Deucher <alexander.deucher@amd.com>
> 
> Adjust the FP handling to avoid nested calls.
> 
> The nested calls cause the warning below
> WARNING: CPU: 3 PID: 384 at arch/x86/kernel/fpu/core.c:129 kernel_fpu_begin
> 
> Fixes: 26803606c5d6 ("drm/amdgpu/display: FP fixes for DCN3.x (v4)")
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> Signed-off-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>

I guess dropping the noinline is fine if we're just calling it via the 
function pointer.

Regards,
Nicholas Kazlauskas

> ---
>   .../drm/amd/display/dc/dcn30/dcn30_resource.c | 43 +++----------------
>   1 file changed, 6 insertions(+), 37 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
> index b379057e669c..d5c81ad55045 100644
> --- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
> +++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
> @@ -1470,20 +1470,8 @@ int dcn30_populate_dml_pipes_from_context(
>   	return pipe_cnt;
>   }
>   
> -/*
> - * This must be noinline to ensure anything that deals with FP registers
> - * is contained within this call; previously our compiling with hard-float
> - * would result in fp instructions being emitted outside of the boundaries
> - * of the DC_FP_START/END macros, which makes sense as the compiler has no
> - * idea about what is wrapped and what is not
> - *
> - * This is largely just a workaround to avoid breakage introduced with 5.6,
> - * ideally all fp-using code should be moved into its own file, only that
> - * should be compiled with hard-float, and all code exported from there
> - * should be strictly wrapped with DC_FP_START/END
> - */
> -static noinline void dcn30_populate_dml_writeback_from_context_fp(
> -		struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
> +void dcn30_populate_dml_writeback_from_context(
> +	struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
>   {
>   	int pipe_cnt, i, j;
>   	double max_calc_writeback_dispclk;
> @@ -1571,14 +1559,6 @@ static noinline void dcn30_populate_dml_writeback_from_context_fp(
>   
>   }
>   
> -void dcn30_populate_dml_writeback_from_context(
> -		struct dc *dc, struct resource_context *res_ctx, display_e2e_pipe_params_st *pipes)
> -{
> -	DC_FP_START();
> -	dcn30_populate_dml_writeback_from_context_fp(dc, res_ctx, pipes);
> -	DC_FP_END();
> -}
> -
>   unsigned int dcn30_calc_max_scaled_time(
>   		unsigned int time_per_pixel,
>   		enum mmhubbub_wbif_mode mode,
> @@ -1977,7 +1957,7 @@ static struct pipe_ctx *dcn30_find_split_pipe(
>   	return pipe;
>   }
>   
> -static bool dcn30_internal_validate_bw(
> +static noinline bool dcn30_internal_validate_bw(
>   		struct dc *dc,
>   		struct dc_state *context,
>   		display_e2e_pipe_params_st *pipes,
> @@ -1999,6 +1979,7 @@ static bool dcn30_internal_validate_bw(
>   
>   	pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate);
>   
> +	DC_FP_START();
>   	if (!pipe_cnt) {
>   		out = true;
>   		goto validate_out;
> @@ -2222,6 +2203,7 @@ static bool dcn30_internal_validate_bw(
>   	out = false;
>   
>   validate_out:
> +	DC_FP_END();
>   	return out;
>   }
>   
> @@ -2404,7 +2386,7 @@ void dcn30_calculate_wm_and_dlg(
>   	DC_FP_END();
>   }
>   
> -static noinline bool dcn30_validate_bandwidth_fp(struct dc *dc,
> +bool dcn30_validate_bandwidth(struct dc *dc,
>   		struct dc_state *context,
>   		bool fast_validate)
>   {
> @@ -2455,19 +2437,6 @@ static noinline bool dcn30_validate_bandwidth_fp(struct dc *dc,
>   	return out;
>   }
>   
> -bool dcn30_validate_bandwidth(struct dc *dc,
> -		struct dc_state *context,
> -		bool fast_validate)
> -{
> -	bool out;
> -
> -	DC_FP_START();
> -	out = dcn30_validate_bandwidth_fp(dc, context, fast_validate);
> -	DC_FP_END();
> -
> -	return out;
> -}
> -
>   static noinline void get_optimal_dcfclk_fclk_for_uclk(unsigned int uclk_mts,
>                                                          unsigned int *optimal_dcfclk,
>                                                          unsigned int *optimal_fclk)
> 

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx