1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
|
From 3dbf2c2bb6a87c4890d500e7b883c8ac91a597b1 Mon Sep 17 00:00:00 2001
From: Jun Lei <Jun.Lei@amd.com>
Date: Thu, 9 May 2019 15:32:27 -0400
Subject: [PATCH 2461/2940] drm/amd/display: update calculated bounding box
logic for NV
[why]
Current calculation of bounding box will cause DML to increase voltage
state due to DPP or DISPCLK, this is unnecessary since from DML perspective
we can max DPP/DISP can be supported at DPM0. This is because
increasing voltage for DPP/DISP is done separately via actual minimum values
of DISP and DPP CLK
[how]
For each calculated state, DPP, DISP, PHY, and DSC clk should always be set to
maximum. FCLK, SOCCLK, and DCFCLK should be based of UCLK.
Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
.../drm/amd/display/dc/dcn20/dcn20_resource.c | 55 ++++++++-----------
1 file changed, 24 insertions(+), 31 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
index d0c279ab0af1..c4fa0b9e7138 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
@@ -2402,54 +2402,47 @@ static void update_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_
struct pp_smu_nv_clock_table *max_clocks, unsigned int *uclk_states, unsigned int num_states)
{
struct _vcs_dpi_voltage_scaling_st calculated_states[MAX_CLOCK_LIMIT_STATES] = {0};
- int i, j;
+ int i;
int num_calculated_states = 0;
+ int min_dcfclk = 0;
if (num_states == 0)
return;
+ if (dc->bb_overrides.min_dcfclk_mhz > 0)
+ min_dcfclk = dc->bb_overrides.min_dcfclk_mhz;
+
for (i = 0; i < num_states; i++) {
- // Find lowest pre-silicon DPM that has equal or higher uCLK
- for (j = 0; j < bb->num_states; j++) {
- if (bb->clock_limits[j].dram_speed_mts * 1000 / 16 >= uclk_states[i])
- break;
- }
+ int min_fclk_required_by_uclk;
+ calculated_states[i].state = i;
+ calculated_states[i].dram_speed_mts = uclk_states[i] * 16 / 1000;
- // If for some reason the available uCLK is higher than all pre-silicon'
- // DPM targets, then we just use the highest one
- if (j >= bb->num_states)
- j = bb->num_states;
+ min_fclk_required_by_uclk = ((unsigned long long)uclk_states[i]) * 1008 / 1000000;
- // Copy that state
- memcpy(&calculated_states[num_calculated_states], &bb->clock_limits[j],
- sizeof(calculated_states[num_calculated_states]));
+ calculated_states[i].fabricclk_mhz = (min_fclk_required_by_uclk < min_dcfclk) ?
+ min_dcfclk : min_fclk_required_by_uclk;
- // Cap uClk to actual
- calculated_states[num_calculated_states].dram_speed_mts = uclk_states[i] * 16 / 1000;
- // Phy clock can be set to max for all states, since there's nothing to optimize
- // for spreadsheet and we request voltage for phy clock by frequency anyway
- calculated_states[num_calculated_states].phyclk_mhz = max_clocks->phyClockInKhz / 1000;
+ calculated_states[i].socclk_mhz = (calculated_states[i].fabricclk_mhz > max_clocks->socClockInKhz / 1000) ?
+ max_clocks->socClockInKhz / 1000 : calculated_states[i].fabricclk_mhz;
- calculated_states[num_calculated_states].state = num_calculated_states;
+ calculated_states[i].dcfclk_mhz = (calculated_states[i].fabricclk_mhz > max_clocks->dcfClockInKhz / 1000) ?
+ max_clocks->dcfClockInKhz / 1000 : calculated_states[i].fabricclk_mhz;
- num_calculated_states++;
- }
+ calculated_states[i].dispclk_mhz = max_clocks->displayClockInKhz / 1000;
+ calculated_states[i].dppclk_mhz = max_clocks->displayClockInKhz / 1000;
+ calculated_states[i].dscclk_mhz = max_clocks->displayClockInKhz / (1000 * 3);
- if (max_clocks->dcfClockInKhz > 0)
- calculated_states[num_calculated_states - 1].dcfclk_mhz = max_clocks->dcfClockInKhz / 1000;
+ calculated_states[i].phyclk_mhz = max_clocks->phyClockInKhz / 1000;
- if (max_clocks->displayClockInKhz > 0) {
- calculated_states[num_calculated_states - 1].dispclk_mhz = max_clocks->displayClockInKhz / 1000;
- calculated_states[num_calculated_states - 1].dppclk_mhz = max_clocks->displayClockInKhz / 1000;
- // DSC always runs at 1/3 of disp clock
- calculated_states[num_calculated_states - 1].dscclk_mhz = max_clocks->displayClockInKhz / (1000 * 3);
+ num_calculated_states++;
}
- if (max_clocks->socClockInKhz > 0)
- calculated_states[num_calculated_states - 1].socclk_mhz = max_clocks->socClockInKhz / 1000;
-
memcpy(bb->clock_limits, calculated_states, sizeof(bb->clock_limits));
bb->num_states = num_calculated_states;
+
+ // Duplicate the last state, DML always an extra state identical to max state to work
+ memcpy(&bb->clock_limits[num_calculated_states], &bb->clock_limits[num_calculated_states - 1], sizeof(struct _vcs_dpi_voltage_scaling_st));
+ bb->clock_limits[num_calculated_states].state = bb->num_states;
}
static void patch_bounding_box(struct dc *dc, struct _vcs_dpi_soc_bounding_box_st *bb)
--
2.17.1
|