gpu: nvgpu: Use gradual slowdown

Program clock slowdown to happen using gradual slowdown. It is
significantly faster than the default slowdown.

Change-Id: I9e5259889637fce2c0b083a424b54af12bb45c25
Signed-off-by: Terje Bergstrom <tbergstrom@nvidia.com>
Reviewed-on: http://git-master/r/819698

3 files changed, 257 insertions, 2 deletions

diff --git a/drivers/gpu/nvgpu/gk20a/hw_therm_gk20a.h b/drivers/gpu/nvgpu/gk20a/hw_therm_gk20a.h
index 42a31c5d..67f1bbc2 100644
--- a/drivers/gpu/nvgpu/gk20a/hw_therm_gk20a.h
+++ b/drivers/gpu/nvgpu/gk20a/hw_therm_gk20a.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2012-2014, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2012-2015, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -74,6 +74,18 @@ static inline u32 therm_config1_r(void)
 {
         return 0x00020050;
 }
+static inline u32 therm_config2_r(void)
+{
+        return 0x00020130;
+}
+static inline u32 therm_config2_slowdown_factor_extended_f(u32 v)
+{
+        return (v & 0x1) << 24;
+}
+static inline u32 therm_config2_grad_enable_f(u32 v)
+{
+        return (v & 0x1) << 31;
+}
 static inline u32 therm_gate_ctrl_r(u32 i)
 {
         return 0x00020200 + i*4;
@@ -182,4 +194,112 @@ static inline u32 therm_clk_slowdown_idle_factor_disabled_f(void)
 {
         return 0x0;
 }
+static inline u32 therm_grad_stepping_table_r(u32 i)
+{
+        return 0x000202c8 + i*4;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_f(u32 v)
+{
+        return (v & 0x3f) << 0;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_m(void)
+{
+        return 0x3f << 0;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by1p5_f(void)
+{
+        return 0x1;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by2_f(void)
+{
+        return 0x2;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by4_f(void)
+{
+        return 0x6;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f(void)
+{
+        return 0xe;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor1_f(u32 v)
+{
+        return (v & 0x3f) << 6;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor1_m(void)
+{
+        return 0x3f << 6;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor2_f(u32 v)
+{
+        return (v & 0x3f) << 12;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor2_m(void)
+{
+        return 0x3f << 12;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor3_f(u32 v)
+{
+        return (v & 0x3f) << 18;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor3_m(void)
+{
+        return 0x3f << 18;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor4_f(u32 v)
+{
+        return (v & 0x3f) << 24;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor4_m(void)
+{
+        return 0x3f << 24;
+}
+static inline u32 therm_grad_stepping0_r(void)
+{
+        return 0x000202c0;
+}
+static inline u32 therm_grad_stepping0_feature_s(void)
+{
+        return 1;
+}
+static inline u32 therm_grad_stepping0_feature_f(u32 v)
+{
+        return (v & 0x1) << 0;
+}
+static inline u32 therm_grad_stepping0_feature_m(void)
+{
+        return 0x1 << 0;
+}
+static inline u32 therm_grad_stepping0_feature_v(u32 r)
+{
+        return (r >> 0) & 0x1;
+}
+static inline u32 therm_grad_stepping0_feature_enable_f(void)
+{
+        return 0x1;
+}
+static inline u32 therm_grad_stepping1_r(void)
+{
+        return 0x000202c4;
+}
+static inline u32 therm_grad_stepping1_pdiv_duration_f(u32 v)
+{
+        return (v & 0x1ffff) << 0;
+}
+static inline u32 therm_clk_timing_r(u32 i)
+{
+        return 0x000203c0 + i*4;
+}
+static inline u32 therm_clk_timing_grad_slowdown_f(u32 v)
+{
+        return (v & 0x1) << 16;
+}
+static inline u32 therm_clk_timing_grad_slowdown_m(void)
+{
+        return 0x1 << 16;
+}
+static inline u32 therm_clk_timing_grad_slowdown_enabled_f(void)
+{
+        return 0x10000;
+}
 #endif
diff --git a/drivers/gpu/nvgpu/gk20a/therm_gk20a.c b/drivers/gpu/nvgpu/gk20a/therm_gk20a.c
index b02113ad..fd4149ae 100644
--- a/drivers/gpu/nvgpu/gk20a/therm_gk20a.c
+++ b/drivers/gpu/nvgpu/gk20a/therm_gk20a.c
@@ -35,6 +35,8 @@ static int gk20a_init_therm_setup_sw(struct gk20a *g)
 
 static int gk20a_init_therm_setup_hw(struct gk20a *g)
 {
+        u32 v;
+
         /* program NV_THERM registers */
         gk20a_writel(g, therm_use_a_r(), NV_THERM_USE_A_INIT);
         gk20a_writel(g, therm_evt_ext_therm_0_r(),
@@ -44,6 +46,35 @@ static int gk20a_init_therm_setup_hw(struct gk20a *g)
         gk20a_writel(g, therm_evt_ext_therm_2_r(),
                 NV_THERM_EVT_EXT_THERM_2_INIT);
 
+        gk20a_writel(g, therm_grad_stepping_table_r(0),
+                therm_grad_stepping_table_slowdown_factor0_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by1p5_f()) |
+                therm_grad_stepping_table_slowdown_factor1_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by2_f()) |
+                therm_grad_stepping_table_slowdown_factor2_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by4_f()) |
+                therm_grad_stepping_table_slowdown_factor3_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()) |
+                therm_grad_stepping_table_slowdown_factor4_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()));
+        gk20a_writel(g, therm_grad_stepping_table_r(1),
+                therm_grad_stepping_table_slowdown_factor0_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()) |
+                therm_grad_stepping_table_slowdown_factor1_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()) |
+                therm_grad_stepping_table_slowdown_factor2_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()) |
+                therm_grad_stepping_table_slowdown_factor3_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()) |
+                therm_grad_stepping_table_slowdown_factor4_f(therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f()));
+
+        v = gk20a_readl(g, therm_clk_timing_r(0));
+        v |= therm_clk_timing_grad_slowdown_enabled_f();
+        gk20a_writel(g, therm_clk_timing_r(0), v);
+
+        v = gk20a_readl(g, therm_config2_r());
+        v |= therm_config2_grad_enable_f(1);
+        v |= therm_config2_slowdown_factor_extended_f(1);
+        gk20a_writel(g, therm_config2_r(), v);
+
+        gk20a_writel(g, therm_grad_stepping1_r(),
+                        therm_grad_stepping1_pdiv_duration_f(32));
+
+        v = gk20a_readl(g, therm_grad_stepping0_r());
+        v |= therm_grad_stepping0_feature_enable_f();
+        gk20a_writel(g, therm_grad_stepping0_r(), v);
+
         return 0;
 }
 
diff --git a/drivers/gpu/nvgpu/gm20b/hw_therm_gm20b.h b/drivers/gpu/nvgpu/gm20b/hw_therm_gm20b.h
index d928d70e..9943cf31 100644
--- a/drivers/gpu/nvgpu/gm20b/hw_therm_gm20b.h
+++ b/drivers/gpu/nvgpu/gm20b/hw_therm_gm20b.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2014, NVIDIA CORPORATION.  All rights reserved.
+ * Copyright (c) 2014-2015, NVIDIA CORPORATION.  All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms and conditions of the GNU General Public License,
@@ -74,6 +74,18 @@ static inline u32 therm_config1_r(void)
 {
         return 0x00020050;
 }
+static inline u32 therm_config2_r(void)
+{
+        return 0x00020130;
+}
+static inline u32 therm_config2_slowdown_factor_extended_f(u32 v)
+{
+        return (v & 0x1) << 24;
+}
+static inline u32 therm_config2_grad_enable_f(u32 v)
+{
+        return (v & 0x1) << 31;
+}
 static inline u32 therm_gate_ctrl_r(u32 i)
 {
         return 0x00020200 + i*4;
@@ -182,4 +194,96 @@ static inline u32 therm_clk_slowdown_idle_factor_disabled_f(void)
 {
         return 0x0;
 }
+static inline u32 therm_grad_stepping_table_r(u32 i)
+{
+        return 0x000202c8 + i*4;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_f(u32 v)
+{
+        return (v & 0x3f) << 0;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_m(void)
+{
+        return 0x3f << 0;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by1p5_f(void)
+{
+        return 0x1;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by2_f(void)
+{
+        return 0x2;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by4_f(void)
+{
+        return 0x6;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor0_fpdiv_by8_f(void)
+{
+        return 0xe;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor1_f(u32 v)
+{
+        return (v & 0x3f) << 6;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor2_f(u32 v)
+{
+        return (v & 0x3f) << 12;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor3_f(u32 v)
+{
+        return (v & 0x3f) << 18;
+}
+static inline u32 therm_grad_stepping_table_slowdown_factor4_f(u32 v)
+{
+        return (v & 0x3f) << 24;
+}
+static inline u32 therm_grad_stepping0_r(void)
+{
+        return 0x000202c0;
+}
+static inline u32 therm_grad_stepping0_feature_s(void)
+{
+        return 1;
+}
+static inline u32 therm_grad_stepping0_feature_f(u32 v)
+{
+        return (v & 0x1) << 0;
+}
+static inline u32 therm_grad_stepping0_feature_m(void)
+{
+        return 0x1 << 0;
+}
+static inline u32 therm_grad_stepping0_feature_v(u32 r)
+{
+        return (r >> 0) & 0x1;
+}
+static inline u32 therm_grad_stepping0_feature_enable_f(void)
+{
+        return 0x1;
+}
+static inline u32 therm_grad_stepping1_r(void)
+{
+        return 0x000202c4;
+}
+static inline u32 therm_grad_stepping1_pdiv_duration_f(u32 v)
+{
+        return (v & 0x1ffff) << 0;
+}
+static inline u32 therm_clk_timing_r(u32 i)
+{
+        return 0x000203c0 + i*4;
+}
+static inline u32 therm_clk_timing_grad_slowdown_f(u32 v)
+{
+        return (v & 0x1) << 16;
+}
+static inline u32 therm_clk_timing_grad_slowdown_m(void)
+{
+        return 0x1 << 16;
+}
+static inline u32 therm_clk_timing_grad_slowdown_enabled_f(void)
+{
+        return 0x10000;
+}
 #endif