1 /* 2 * Copyright 2016 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 #ifndef __AMDGPU_XGMI_H__ 23 #define __AMDGPU_XGMI_H__ 24 25 #include <drm/task_barrier.h> 26 #include "amdgpu_psp.h" 27 #include "amdgpu_ras.h" 28 29 struct amdgpu_hive_info { 30 struct kobject kobj; 31 uint64_t hive_id; 32 struct list_head device_list; 33 struct list_head node; 34 atomic_t number_devices; 35 struct mutex hive_lock; 36 int hi_req_count; 37 struct amdgpu_device *hi_req_gpu; 38 struct task_barrier tb; 39 enum { 40 AMDGPU_XGMI_PSTATE_MIN, 41 AMDGPU_XGMI_PSTATE_MAX_VEGA20, 42 AMDGPU_XGMI_PSTATE_UNKNOWN 43 } pstate; 44 45 struct amdgpu_reset_domain *reset_domain; 46 atomic_t ras_recovery; 47 struct ras_event_manager event_mgr; 48 struct work_struct reset_on_init_work; 49 atomic_t requested_nps_mode; 50 }; 51 52 struct amdgpu_pcs_ras_field { 53 const char *err_name; 54 uint32_t pcs_err_mask; 55 uint32_t pcs_err_shift; 56 }; 57 58 /** 59 * Bandwidth range reporting comes in two modes. 60 * 61 * PER_LINK - range for any xgmi link 62 * PER_PEER - range of max of single xgmi link to max of multiple links based on source peer 63 */ 64 enum amdgpu_xgmi_bw_mode { 65 AMDGPU_XGMI_BW_MODE_PER_LINK = 0, 66 AMDGPU_XGMI_BW_MODE_PER_PEER 67 }; 68 69 enum amdgpu_xgmi_bw_unit { 70 AMDGPU_XGMI_BW_UNIT_GBYTES = 0, 71 AMDGPU_XGMI_BW_UNIT_MBYTES 72 }; 73 74 extern struct amdgpu_xgmi_ras xgmi_ras; 75 struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev); 76 void amdgpu_put_xgmi_hive(struct amdgpu_hive_info *hive); 77 int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev); 78 int amdgpu_xgmi_add_device(struct amdgpu_device *adev); 79 int amdgpu_xgmi_remove_device(struct amdgpu_device *adev); 80 int amdgpu_xgmi_set_pstate(struct amdgpu_device *adev, int pstate); 81 int amdgpu_xgmi_get_hops_count(struct amdgpu_device *adev, struct amdgpu_device *peer_adev); 82 int amdgpu_xgmi_get_bandwidth(struct amdgpu_device *adev, struct amdgpu_device *peer_adev, 83 enum amdgpu_xgmi_bw_mode bw_mode, enum amdgpu_xgmi_bw_unit bw_unit, 84 uint32_t *min_bw, uint32_t *max_bw); 85 bool amdgpu_xgmi_get_is_sharing_enabled(struct amdgpu_device *adev, 86 struct amdgpu_device *peer_adev); 87 uint64_t amdgpu_xgmi_get_relative_phy_addr(struct amdgpu_device *adev, 88 uint64_t addr); 89 static inline bool amdgpu_xgmi_same_hive(struct amdgpu_device *adev, 90 struct amdgpu_device *bo_adev) 91 { 92 return (amdgpu_use_xgmi_p2p && 93 adev != bo_adev && 94 adev->gmc.xgmi.hive_id && 95 adev->gmc.xgmi.hive_id == bo_adev->gmc.xgmi.hive_id); 96 } 97 int amdgpu_xgmi_ras_sw_init(struct amdgpu_device *adev); 98 int amdgpu_xgmi_reset_on_init(struct amdgpu_device *adev); 99 100 int amdgpu_xgmi_request_nps_change(struct amdgpu_device *adev, 101 struct amdgpu_hive_info *hive, 102 int req_nps_mode); 103 int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev, 104 int global_link_num); 105 106 #endif 107