1 /* 2 * Copyright 2019 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_ras.h" 25 26 int amdgpu_umc_ras_late_init(struct amdgpu_device *adev) 27 { 28 int r; 29 struct ras_fs_if fs_info = { 30 .sysfs_name = "umc_err_count", 31 }; 32 struct ras_ih_if ih_info = { 33 .cb = amdgpu_umc_process_ras_data_cb, 34 }; 35 36 if (!adev->umc.ras_if) { 37 adev->umc.ras_if = 38 kmalloc(sizeof(struct ras_common_if), GFP_KERNEL); 39 if (!adev->umc.ras_if) 40 return -ENOMEM; 41 adev->umc.ras_if->block = AMDGPU_RAS_BLOCK__UMC; 42 adev->umc.ras_if->type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE; 43 adev->umc.ras_if->sub_block_index = 0; 44 strcpy(adev->umc.ras_if->name, "umc"); 45 } 46 ih_info.head = fs_info.head = *adev->umc.ras_if; 47 48 r = amdgpu_ras_late_init(adev, adev->umc.ras_if, 49 &fs_info, &ih_info); 50 if (r) 51 goto free; 52 53 if (amdgpu_ras_is_supported(adev, adev->umc.ras_if->block)) { 54 r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0); 55 if (r) 56 goto late_fini; 57 } else { 58 r = 0; 59 goto free; 60 } 61 62 /* ras init of specific umc version */ 63 if (adev->umc.ras_funcs && 64 adev->umc.ras_funcs->err_cnt_init) 65 adev->umc.ras_funcs->err_cnt_init(adev); 66 67 return 0; 68 69 late_fini: 70 amdgpu_ras_late_fini(adev, adev->umc.ras_if, &ih_info); 71 free: 72 kfree(adev->umc.ras_if); 73 adev->umc.ras_if = NULL; 74 return r; 75 } 76 77 void amdgpu_umc_ras_fini(struct amdgpu_device *adev) 78 { 79 if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__UMC) && 80 adev->umc.ras_if) { 81 struct ras_common_if *ras_if = adev->umc.ras_if; 82 struct ras_ih_if ih_info = { 83 .head = *ras_if, 84 .cb = amdgpu_umc_process_ras_data_cb, 85 }; 86 87 amdgpu_ras_late_fini(adev, ras_if, &ih_info); 88 kfree(ras_if); 89 } 90 } 91 92 int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, 93 void *ras_error_status, 94 struct amdgpu_iv_entry *entry) 95 { 96 struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; 97 struct amdgpu_ras *con = amdgpu_ras_get_context(adev); 98 99 kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); 100 if (adev->umc.ras_funcs && 101 adev->umc.ras_funcs->query_ras_error_count) 102 adev->umc.ras_funcs->query_ras_error_count(adev, ras_error_status); 103 104 if (adev->umc.ras_funcs && 105 adev->umc.ras_funcs->query_ras_error_address && 106 adev->umc.max_ras_err_cnt_per_query) { 107 err_data->err_addr = 108 kcalloc(adev->umc.max_ras_err_cnt_per_query, 109 sizeof(struct eeprom_table_record), GFP_KERNEL); 110 111 /* still call query_ras_error_address to clear error status 112 * even NOMEM error is encountered 113 */ 114 if(!err_data->err_addr) 115 dev_warn(adev->dev, "Failed to alloc memory for " 116 "umc error address record!\n"); 117 118 /* umc query_ras_error_address is also responsible for clearing 119 * error status 120 */ 121 adev->umc.ras_funcs->query_ras_error_address(adev, ras_error_status); 122 } 123 124 /* only uncorrectable error needs gpu reset */ 125 if (err_data->ue_count) { 126 dev_info(adev->dev, "%ld uncorrectable hardware errors " 127 "detected in UMC block\n", 128 err_data->ue_count); 129 130 if ((amdgpu_bad_page_threshold != 0) && 131 err_data->err_addr_cnt) { 132 amdgpu_ras_add_bad_pages(adev, err_data->err_addr, 133 err_data->err_addr_cnt); 134 amdgpu_ras_save_bad_pages(adev); 135 136 if (adev->smu.ppt_funcs && adev->smu.ppt_funcs->send_hbm_bad_pages_num) 137 adev->smu.ppt_funcs->send_hbm_bad_pages_num(&adev->smu, con->eeprom_control.num_recs); 138 } 139 140 amdgpu_ras_reset_gpu(adev); 141 } 142 143 kfree(err_data->err_addr); 144 return AMDGPU_RAS_SUCCESS; 145 } 146 147 int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, 148 struct amdgpu_irq_src *source, 149 struct amdgpu_iv_entry *entry) 150 { 151 struct ras_common_if *ras_if = adev->umc.ras_if; 152 struct ras_dispatch_if ih_data = { 153 .entry = entry, 154 }; 155 156 if (!ras_if) 157 return 0; 158 159 ih_data.head = *ras_if; 160 161 amdgpu_ras_interrupt_dispatch(adev, &ih_data); 162 return 0; 163 } 164