Line data Source code
1 : /*
2 : * Copyright 2018 Advanced Micro Devices, Inc.
3 : *
4 : * Permission is hereby granted, free of charge, to any person obtaining a
5 : * copy of this software and associated documentation files (the "Software"),
6 : * to deal in the Software without restriction, including without limitation
7 : * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 : * and/or sell copies of the Software, and to permit persons to whom the
9 : * Software is furnished to do so, subject to the following conditions:
10 : *
11 : * The above copyright notice and this permission notice shall be included in
12 : * all copies or substantial portions of the Software.
13 : *
14 : * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15 : * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 : * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
17 : * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
18 : * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
19 : * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
20 : * OTHER DEALINGS IN THE SOFTWARE.
21 : *
22 : */
23 :
24 : #include "amdgpu.h"
25 : #include "amdgpu_sdma.h"
26 : #include "amdgpu_ras.h"
27 :
28 : #define AMDGPU_CSA_SDMA_SIZE 64
29 : /* SDMA CSA reside in the 3rd page of CSA */
30 : #define AMDGPU_CSA_SDMA_OFFSET (4096 * 2)
31 :
32 : /*
33 : * GPU SDMA IP block helpers function.
34 : */
35 :
36 0 : struct amdgpu_sdma_instance *amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring)
37 : {
38 0 : struct amdgpu_device *adev = ring->adev;
39 : int i;
40 :
41 0 : for (i = 0; i < adev->sdma.num_instances; i++)
42 0 : if (ring == &adev->sdma.instance[i].ring ||
43 0 : ring == &adev->sdma.instance[i].page)
44 0 : return &adev->sdma.instance[i];
45 :
46 : return NULL;
47 : }
48 :
49 0 : int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index)
50 : {
51 0 : struct amdgpu_device *adev = ring->adev;
52 : int i;
53 :
54 0 : for (i = 0; i < adev->sdma.num_instances; i++) {
55 0 : if (ring == &adev->sdma.instance[i].ring ||
56 0 : ring == &adev->sdma.instance[i].page) {
57 0 : *index = i;
58 0 : return 0;
59 : }
60 : }
61 :
62 : return -EINVAL;
63 : }
64 :
65 0 : uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring,
66 : unsigned vmid)
67 : {
68 0 : struct amdgpu_device *adev = ring->adev;
69 : uint64_t csa_mc_addr;
70 0 : uint32_t index = 0;
71 : int r;
72 :
73 : /* don't enable OS preemption on SDMA under SRIOV */
74 0 : if (amdgpu_sriov_vf(adev) || vmid == 0 || !amdgpu_mcbp)
75 : return 0;
76 :
77 0 : if (ring->is_mes_queue) {
78 0 : uint32_t offset = 0;
79 :
80 0 : offset = offsetof(struct amdgpu_mes_ctx_meta_data,
81 : sdma[ring->idx].sdma_meta_data);
82 0 : csa_mc_addr = amdgpu_mes_ctx_get_offs_gpu_addr(ring, offset);
83 : } else {
84 0 : r = amdgpu_sdma_get_index_from_ring(ring, &index);
85 :
86 0 : if (r || index > 31)
87 : csa_mc_addr = 0;
88 : else
89 0 : csa_mc_addr = amdgpu_csa_vaddr(adev) +
90 0 : AMDGPU_CSA_SDMA_OFFSET +
91 0 : index * AMDGPU_CSA_SDMA_SIZE;
92 : }
93 :
94 : return csa_mc_addr;
95 : }
96 :
97 0 : int amdgpu_sdma_ras_late_init(struct amdgpu_device *adev,
98 : struct ras_common_if *ras_block)
99 : {
100 : int r, i;
101 :
102 0 : r = amdgpu_ras_block_late_init(adev, ras_block);
103 0 : if (r)
104 : return r;
105 :
106 0 : if (amdgpu_ras_is_supported(adev, ras_block->block)) {
107 0 : for (i = 0; i < adev->sdma.num_instances; i++) {
108 0 : r = amdgpu_irq_get(adev, &adev->sdma.ecc_irq,
109 : AMDGPU_SDMA_IRQ_INSTANCE0 + i);
110 0 : if (r)
111 : goto late_fini;
112 : }
113 : }
114 :
115 : return 0;
116 :
117 : late_fini:
118 0 : amdgpu_ras_block_late_fini(adev, ras_block);
119 0 : return r;
120 : }
121 :
122 0 : int amdgpu_sdma_process_ras_data_cb(struct amdgpu_device *adev,
123 : void *err_data,
124 : struct amdgpu_iv_entry *entry)
125 : {
126 0 : kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
127 :
128 0 : if (amdgpu_sriov_vf(adev))
129 : return AMDGPU_RAS_SUCCESS;
130 :
131 0 : amdgpu_ras_reset_gpu(adev);
132 :
133 0 : return AMDGPU_RAS_SUCCESS;
134 : }
135 :
136 0 : int amdgpu_sdma_process_ecc_irq(struct amdgpu_device *adev,
137 : struct amdgpu_irq_src *source,
138 : struct amdgpu_iv_entry *entry)
139 : {
140 0 : struct ras_common_if *ras_if = adev->sdma.ras_if;
141 0 : struct ras_dispatch_if ih_data = {
142 : .entry = entry,
143 : };
144 :
145 0 : if (!ras_if)
146 : return 0;
147 :
148 0 : ih_data.head = *ras_if;
149 :
150 0 : amdgpu_ras_interrupt_dispatch(adev, &ih_data);
151 0 : return 0;
152 : }
|