1 /*
2  * Copyright 2021 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 #ifndef __AMDGPU_RESET_H__
25 #define __AMDGPU_RESET_H__
26 
27 #include "amdgpu.h"
28 
29 enum AMDGPU_RESET_FLAGS {
30 
31 	AMDGPU_NEED_FULL_RESET = 0,
32 	AMDGPU_SKIP_HW_RESET = 1,
33 	AMDGPU_RESET_FOR_DEVICE_REMOVE = 2,
34 };
35 
36 struct amdgpu_reset_context {
37 	enum amd_reset_method method;
38 	struct amdgpu_device *reset_req_dev;
39 	struct amdgpu_job *job;
40 	struct amdgpu_hive_info *hive;
41 	struct list_head *reset_device_list;
42 	unsigned long flags;
43 };
44 
45 struct amdgpu_reset_handler {
46 	enum amd_reset_method reset_method;
47 	struct list_head handler_list;
48 	int (*prepare_env)(struct amdgpu_reset_control *reset_ctl,
49 			   struct amdgpu_reset_context *context);
50 	int (*prepare_hwcontext)(struct amdgpu_reset_control *reset_ctl,
51 				 struct amdgpu_reset_context *context);
52 	int (*perform_reset)(struct amdgpu_reset_control *reset_ctl,
53 			     struct amdgpu_reset_context *context);
54 	int (*restore_hwcontext)(struct amdgpu_reset_control *reset_ctl,
55 				 struct amdgpu_reset_context *context);
56 	int (*restore_env)(struct amdgpu_reset_control *reset_ctl,
57 			   struct amdgpu_reset_context *context);
58 
59 	int (*do_reset)(struct amdgpu_device *adev);
60 };
61 
62 struct amdgpu_reset_control {
63 	void *handle;
64 	struct work_struct reset_work;
65 	struct mutex reset_lock;
66 	struct list_head reset_handlers;
67 	atomic_t in_reset;
68 	enum amd_reset_method active_reset;
69 	struct amdgpu_reset_handler *(*get_reset_handler)(
70 		struct amdgpu_reset_control *reset_ctl,
71 		struct amdgpu_reset_context *context);
72 	void (*async_reset)(struct work_struct *work);
73 };
74 
75 
76 enum amdgpu_reset_domain_type {
77 	SINGLE_DEVICE,
78 	XGMI_HIVE
79 };
80 
81 struct amdgpu_reset_domain {
82 	struct kref refcount;
83 	struct workqueue_struct *wq;
84 	enum amdgpu_reset_domain_type type;
85 	struct rw_semaphore sem;
86 	atomic_t in_gpu_reset;
87 	atomic_t reset_res;
88 };
89 
90 
91 int amdgpu_reset_init(struct amdgpu_device *adev);
92 int amdgpu_reset_fini(struct amdgpu_device *adev);
93 
94 int amdgpu_reset_prepare_hwcontext(struct amdgpu_device *adev,
95 				   struct amdgpu_reset_context *reset_context);
96 
97 int amdgpu_reset_perform_reset(struct amdgpu_device *adev,
98 			       struct amdgpu_reset_context *reset_context);
99 
100 int amdgpu_reset_add_handler(struct amdgpu_reset_control *reset_ctl,
101 			     struct amdgpu_reset_handler *handler);
102 
103 struct amdgpu_reset_domain *amdgpu_reset_create_reset_domain(enum amdgpu_reset_domain_type type,
104 							     char *wq_name);
105 
106 void amdgpu_reset_destroy_reset_domain(struct kref *ref);
107 
108 static inline bool amdgpu_reset_get_reset_domain(struct amdgpu_reset_domain *domain)
109 {
110 	return kref_get_unless_zero(&domain->refcount) != 0;
111 }
112 
113 static inline void amdgpu_reset_put_reset_domain(struct amdgpu_reset_domain *domain)
114 {
115 	if (domain)
116 		kref_put(&domain->refcount, amdgpu_reset_destroy_reset_domain);
117 }
118 
119 static inline bool amdgpu_reset_domain_schedule(struct amdgpu_reset_domain *domain,
120 						struct work_struct *work)
121 {
122 	return queue_work(domain->wq, work);
123 }
124 
125 void amdgpu_device_lock_reset_domain(struct amdgpu_reset_domain *reset_domain);
126 
127 void amdgpu_device_unlock_reset_domain(struct amdgpu_reset_domain *reset_domain);
128 
129 #endif
130