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
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
|
From e72c62ba056842dc3823db3053fc073cb56a7dc8 Mon Sep 17 00:00:00 2001
From: Dennis Li <Dennis.Li@amd.com>
Date: Thu, 16 Jan 2020 11:25:08 +0800
Subject: [PATCH 1413/1453] drm/amdgpu: abstract EDC counter clear to a
separated function
1. Add IP prefix for the IP related codes.
2. Refactor the code to clear EDC counter.
Change-Id: I1cd9ec304a7ace9a74480264d24368fd11a87833
Signed-off-by: Dennis Li <Dennis.Li@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
---
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 112 ++++++++++++++++++--------
1 file changed, 77 insertions(+), 35 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 90f0f44d82d9..63f7edb3f9b8 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -736,6 +736,7 @@ static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring);
static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring);
static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
void *ras_error_status);
+static void gfx_v9_0_clear_ras_edc_counter(struct amdgpu_device *adev);
static int gfx_v9_0_ras_error_inject(struct amdgpu_device *adev,
void *inject_if);
@@ -4039,7 +4040,7 @@ static const struct soc15_reg_entry sgpr2_init_regs[] = {
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x0000ff00 },
};
-static const struct soc15_reg_entry sec_ded_counter_registers[] = {
+static const struct soc15_reg_entry gfx_v9_0_edc_counter_regs[] = {
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT), 0, 1, 1},
@@ -4128,7 +4129,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
struct amdgpu_ib ib;
struct dma_fence *f = NULL;
- int r, i, j, k;
+ int r, i;
unsigned total_size, vgpr_offset, sgpr_offset;
u64 gpu_addr;
@@ -4274,18 +4275,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
goto fail;
}
- /* read back registers to clear the counters */
- mutex_lock(&adev->grbm_idx_mutex);
- for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++) {
- for (j = 0; j < sec_ded_counter_registers[i].se_num; j++) {
- for (k = 0; k < sec_ded_counter_registers[i].instance; k++) {
- gfx_v9_0_select_se_sh(adev, j, 0x0, k);
- RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
- }
- }
- }
- WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
- mutex_unlock(&adev->grbm_idx_mutex);
+ gfx_v9_0_clear_ras_edc_counter(adev);
fail:
amdgpu_ib_free(adev, &ib, NULL);
@@ -5564,7 +5554,7 @@ static int gfx_v9_0_priv_inst_irq(struct amdgpu_device *adev,
}
-static const struct soc15_ras_field_entry gc_ras_fields_vg20[] = {
+static const struct soc15_ras_field_entry gfx_v9_0_ras_fields[] = {
{ "CPC_SCRATCH", SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT),
SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, SEC_COUNT),
SOC15_REG_FIELD(CPC_EDC_SCRATCH_CNT, DED_COUNT)
@@ -6137,7 +6127,7 @@ static int gfx_v9_0_query_utc_edc_status(struct amdgpu_device *adev,
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT, 0);
- for (i = 0; i < 16; i++) {
+ for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT);
@@ -6156,7 +6146,7 @@ static int gfx_v9_0_query_utc_edc_status(struct amdgpu_device *adev,
}
}
- for (i = 0; i < 7; i++) {
+ for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT);
@@ -6177,7 +6167,7 @@ static int gfx_v9_0_query_utc_edc_status(struct amdgpu_device *adev,
}
}
- for (i = 0; i < 4; i++) {
+ for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT);
@@ -6189,7 +6179,7 @@ static int gfx_v9_0_query_utc_edc_status(struct amdgpu_device *adev,
}
}
- for (i = 0; i < 32; i++) {
+ for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, i);
data = RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT);
@@ -6216,36 +6206,36 @@ static int gfx_v9_0_query_utc_edc_status(struct amdgpu_device *adev,
return 0;
}
-static int __get_ras_error_count(const struct soc15_reg_entry *reg,
+static int gfx_v9_0_ras_error_count(const struct soc15_reg_entry *reg,
uint32_t se_id, uint32_t inst_id, uint32_t value,
uint32_t *sec_count, uint32_t *ded_count)
{
uint32_t i;
uint32_t sec_cnt, ded_cnt;
- for (i = 0; i < ARRAY_SIZE(gc_ras_fields_vg20); i++) {
- if(gc_ras_fields_vg20[i].reg_offset != reg->reg_offset ||
- gc_ras_fields_vg20[i].seg != reg->seg ||
- gc_ras_fields_vg20[i].inst != reg->inst)
+ for (i = 0; i < ARRAY_SIZE(gfx_v9_0_ras_fields); i++) {
+ if(gfx_v9_0_ras_fields[i].reg_offset != reg->reg_offset ||
+ gfx_v9_0_ras_fields[i].seg != reg->seg ||
+ gfx_v9_0_ras_fields[i].inst != reg->inst)
continue;
sec_cnt = (value &
- gc_ras_fields_vg20[i].sec_count_mask) >>
- gc_ras_fields_vg20[i].sec_count_shift;
+ gfx_v9_0_ras_fields[i].sec_count_mask) >>
+ gfx_v9_0_ras_fields[i].sec_count_shift;
if (sec_cnt) {
DRM_INFO("GFX SubBlock %s, Instance[%d][%d], SEC %d\n",
- gc_ras_fields_vg20[i].name,
+ gfx_v9_0_ras_fields[i].name,
se_id, inst_id,
sec_cnt);
*sec_count += sec_cnt;
}
ded_cnt = (value &
- gc_ras_fields_vg20[i].ded_count_mask) >>
- gc_ras_fields_vg20[i].ded_count_shift;
+ gfx_v9_0_ras_fields[i].ded_count_mask) >>
+ gfx_v9_0_ras_fields[i].ded_count_shift;
if (ded_cnt) {
DRM_INFO("GFX SubBlock %s, Instance[%d][%d], DED %d\n",
- gc_ras_fields_vg20[i].name,
+ gfx_v9_0_ras_fields[i].name,
se_id, inst_id,
ded_cnt);
*ded_count += ded_cnt;
@@ -6255,6 +6245,58 @@ static int __get_ras_error_count(const struct soc15_reg_entry *reg,
return 0;
}
+static void gfx_v9_0_clear_ras_edc_counter(struct amdgpu_device *adev)
+{
+ int i, j, k;
+
+ /* read back registers to clear the counters */
+ mutex_lock(&adev->grbm_idx_mutex);
+ for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {
+ for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) {
+ for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) {
+ gfx_v9_0_select_se_sh(adev, j, 0x0, k);
+ RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i]));
+ }
+ }
+ }
+ WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
+ mutex_unlock(&adev->grbm_idx_mutex);
+
+ WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT, 0);
+ WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT, 0);
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT, 0);
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT, 0);
+
+ for (i = 0; i < ARRAY_SIZE(vml2_mems); i++) {
+ WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, i);
+ RREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_CNT);
+ }
+
+ for (i = 0; i < ARRAY_SIZE(vml2_walker_mems); i++) {
+ WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, i);
+ RREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_CNT);
+ }
+
+ for (i = 0; i < ARRAY_SIZE(atc_l2_cache_2m_mems); i++) {
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, i);
+ RREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_CNT);
+ }
+
+ for (i = 0; i < ARRAY_SIZE(atc_l2_cache_4k_mems); i++) {
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, i);
+ RREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_CNT);
+ }
+
+ WREG32_SOC15(GC, 0, mmVM_L2_MEM_ECC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmVM_L2_WALKER_MEM_ECC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_2M_EDC_INDEX, 255);
+ WREG32_SOC15(GC, 0, mmATC_L2_CACHE_4K_EDC_INDEX, 255);
+}
+
static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
void *ras_error_status)
{
@@ -6271,14 +6313,14 @@ static int gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
mutex_lock(&adev->grbm_idx_mutex);
- for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++) {
- for (j = 0; j < sec_ded_counter_registers[i].se_num; j++) {
- for (k = 0; k < sec_ded_counter_registers[i].instance; k++) {
+ for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {
+ for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) {
+ for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) {
gfx_v9_0_select_se_sh(adev, j, 0, k);
reg_value =
- RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
+ RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i]));
if (reg_value)
- __get_ras_error_count(&sec_ded_counter_registers[i],
+ gfx_v9_0_ras_error_count(&gfx_v9_0_edc_counter_regs[i],
j, k, reg_value,
&sec_count, &ded_count);
}
--
2.17.1
|
