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
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
|
From 9e3086daf194226378bd489d9089328d3cf3cbc9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Christian=20K=C3=B6nig?= <christian.koenig@amd.com>
Date: Fri, 18 Dec 2015 22:13:12 +0100
Subject: [PATCH 0222/1110] drm/amdgpu: keep the prefered/allowed domains in
the BO
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Stop copying that to the bo list entry, it doesn't change anyway.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Kalyan Alle <kalyan.alle@amd.com>
---
drivers/gpu/drm/amd/amdgpu/amdgpu.h | 6 +-
drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c | 14 ++---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 86 ++++++++++++++++++++++-------
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 14 +++--
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 15 +++--
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 4 --
6 files changed, 91 insertions(+), 48 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index b7b5d2e..d5e4503 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -445,8 +445,6 @@ struct amdgpu_bo_list_entry {
struct amdgpu_bo *robj;
struct ttm_validate_buffer tv;
struct amdgpu_bo_va *bo_va;
- unsigned prefered_domains;
- unsigned allowed_domains;
uint32_t priority;
};
@@ -484,6 +482,8 @@ struct amdgpu_bo {
struct list_head list;
/* Protected by tbo.reserved */
u32 initial_domain;
+ u32 prefered_domains;
+ u32 allowed_domains;
struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
struct ttm_placement placement;
struct ttm_buffer_object tbo;
@@ -1064,8 +1064,6 @@ struct amdgpu_bo_list {
struct amdgpu_bo_list *
amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id);
-void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list,
- struct list_head *validated);
void amdgpu_bo_list_put(struct amdgpu_bo_list *list);
void amdgpu_bo_list_free(struct amdgpu_bo_list *list);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 9da4bd0..1c6c0ac 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -111,23 +111,17 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
drm_gem_object_unreference_unlocked(gobj);
entry->priority = min(info[i].bo_priority,
AMDGPU_BO_LIST_MAX_PRIORITY);
- entry->prefered_domains = entry->robj->initial_domain;
- entry->allowed_domains = entry->prefered_domains;
- if (entry->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
- entry->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
- if (amdgpu_ttm_tt_has_userptr(entry->robj->tbo.ttm)) {
+ if (amdgpu_ttm_tt_has_userptr(entry->robj->tbo.ttm))
has_userptr = true;
- entry->prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
- entry->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
}
entry->tv.bo = &entry->robj->tbo;
entry->tv.shared = true;
- if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_GDS)
+ if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_GDS)
gds_obj = entry->robj;
- if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_GWS)
+ if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_GWS)
gws_obj = entry->robj;
- if (entry->prefered_domains == AMDGPU_GEM_DOMAIN_OA)
+ if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_OA)
oa_obj = entry->robj;
trace_amdgpu_bo_list_set(list, entry->robj);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index 76ec99d..dba4bad 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -30,6 +30,47 @@
#include "amdgpu.h"
#include "amdgpu_trace.h"
+#define AMDGPU_CS_MAX_PRIORITY 32u
+#define AMDGPU_CS_NUM_BUCKETS (AMDGPU_CS_MAX_PRIORITY + 1)
+
+/* This is based on the bucket sort with O(n) time complexity.
+ * An item with priority "i" is added to bucket[i]. The lists are then
+ * concatenated in descending order.
+ */
+struct amdgpu_cs_buckets {
+ struct list_head bucket[AMDGPU_CS_NUM_BUCKETS];
+};
+
+static void amdgpu_cs_buckets_init(struct amdgpu_cs_buckets *b)
+{
+ unsigned i;
+
+ for (i = 0; i < AMDGPU_CS_NUM_BUCKETS; i++)
+ INIT_LIST_HEAD(&b->bucket[i]);
+}
+
+static void amdgpu_cs_buckets_add(struct amdgpu_cs_buckets *b,
+ struct list_head *item, unsigned priority)
+{
+ /* Since buffers which appear sooner in the relocation list are
+ * likely to be used more often than buffers which appear later
+ * in the list, the sort mustn't change the ordering of buffers
+ * with the same priority, i.e. it must be stable.
+ */
+ list_add_tail(item, &b->bucket[min(priority, AMDGPU_CS_MAX_PRIORITY)]);
+}
+
+static void amdgpu_cs_buckets_get_list(struct amdgpu_cs_buckets *b,
+ struct list_head *out_list)
+{
+ unsigned i;
+
+ /* Connect the sorted buckets in the output list. */
+ for (i = 0; i < AMDGPU_CS_NUM_BUCKETS; i++) {
+ list_splice(&b->bucket[i], out_list);
+ }
+}
+
int amdgpu_cs_get_ring(struct amdgpu_device *adev, u32 ip_type,
u32 ip_instance, u32 ring,
struct amdgpu_ring **out_ring)
@@ -107,8 +148,6 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
}
p->uf_entry.robj = amdgpu_bo_ref(p->uf.bo);
- p->uf_entry.prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
- p->uf_entry.allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
p->uf_entry.priority = 0;
p->uf_entry.tv.bo = &p->uf_entry.robj->tbo;
p->uf_entry.tv.shared = true;
@@ -139,13 +178,16 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
ret = -EINVAL;
goto free_chunk;
}
+
+ p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
/* get chunks */
+ INIT_LIST_HEAD(&p->validated);
chunk_array_user = (uint64_t __user *)(unsigned long)(cs->in.chunks);
if (copy_from_user(chunk_array, chunk_array_user,
sizeof(uint64_t)*cs->in.num_chunks)) {
ret = -EFAULT;
- goto put_ctx;
+ goto put_bo_list;
}
p->nchunks = cs->in.num_chunks;
@@ -153,7 +195,7 @@ int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
GFP_KERNEL);
if (!p->chunks) {
ret = -ENOMEM;
- goto put_ctx;
+ goto put_bo_list;
}
for (i = 0; i < p->nchunks; i++) {
@@ -229,7 +271,9 @@ free_partial_kdata:
for (; i >= 0; i--)
drm_free_large(p->chunks[i].kdata);
kfree(p->chunks);
-put_ctx:
+put_bo_list:
+ if (p->bo_list)
+ amdgpu_bo_list_put(p->bo_list);
amdgpu_ctx_put(p->ctx);
free_chunk:
kfree(chunk_array);
@@ -315,9 +359,9 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
* completely.
*/
if (p->bytes_moved <= p->bytes_moved_threshold)
- domain = lobj->prefered_domains;
+ domain = bo->prefered_domains;
else
- domain = lobj->allowed_domains;
+ domain = bo->allowed_domains;
retry:
amdgpu_ttm_placement_from_domain(bo, domain);
@@ -327,36 +371,38 @@ int amdgpu_cs_list_validate(struct amdgpu_cs_parser *p,
initial_bytes_moved;
if (unlikely(r)) {
- if (r != -ERESTARTSYS && domain != lobj->allowed_domains) {
- domain = lobj->allowed_domains;
- goto retry;
+ if (r != -ERESTARTSYS && domain != bo->allowed_domains) {
+ domain = bo->allowed_domains;
+ goto retry;
}
+ return r;
}
}
return 0;
}
-static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
- union drm_amdgpu_cs *cs)
+static int amdgpu_cs_parser_relocs(struct amdgpu_cs_parser *p)
{
- struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+ struct amdgpu_cs_buckets buckets;
struct list_head duplicates;
bool need_mmap_lock = false;
- int r;
-
- INIT_LIST_HEAD(&p->validated);
- p->bo_list = amdgpu_bo_list_get(fpriv, cs->in.bo_list_handle);
+ int i, r;
+
if (p->bo_list) {
need_mmap_lock = p->bo_list->has_userptr;
- amdgpu_bo_list_get_list(p->bo_list, &p->validated);
+ amdgpu_cs_buckets_init(&buckets);
+ for (i = 0; i < p->bo_list->num_entries; i++)
+ amdgpu_cs_buckets_add(&buckets, &p->bo_list->array[i].tv.head,
+ p->bo_list->array[i].priority);
+
+ amdgpu_cs_buckets_get_list(&buckets, &p->validated);
}
INIT_LIST_HEAD(&duplicates);
amdgpu_vm_get_pd_bo(&fpriv->vm, &p->validated, &p->vm_pd);
- if (p->uf.bo)
list_add(&p->uf_entry.tv.head, &p->validated);
if (need_mmap_lock)
@@ -774,7 +820,7 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
r = amdgpu_cs_handle_lockup(adev, r);
return r;
}
- r = amdgpu_cs_parser_bos(&parser, data);
+ r = amdgpu_cs_parser_relocs(&parser);
if (r == -ENOMEM)
DRM_ERROR("Not enough memory for command submission!\n");
else if (r && r != -ERESTARTSYS)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index cb34ff6..95e12f4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -252,6 +252,8 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
goto handle_lockup;
bo = gem_to_amdgpu_bo(gobj);
+ bo->prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
+ bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
r = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
if (r)
goto release_object;
@@ -629,7 +631,7 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
info.bo_size = robj->gem_base.size;
info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
- info.domains = robj->initial_domain;
+ info.domains = robj->prefered_domains;
info.domain_flags = robj->flags;
amdgpu_bo_unreserve(robj);
if (copy_to_user(out, &info, sizeof(info)))
@@ -642,9 +644,13 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
amdgpu_bo_unreserve(robj);
break;
}
- robj->initial_domain = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
- AMDGPU_GEM_DOMAIN_GTT |
- AMDGPU_GEM_DOMAIN_CPU);
+ robj->prefered_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
+ AMDGPU_GEM_DOMAIN_GTT |
+ AMDGPU_GEM_DOMAIN_CPU);
+ robj->allowed_domains = robj->prefered_domains;
+ if (robj->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
+ robj->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
+
amdgpu_bo_unreserve(robj);
break;
default:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 73628c7..b79a4f3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -254,12 +254,15 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
bo->adev = adev;
INIT_LIST_HEAD(&bo->list);
INIT_LIST_HEAD(&bo->va);
- bo->initial_domain = domain & (AMDGPU_GEM_DOMAIN_VRAM |
- AMDGPU_GEM_DOMAIN_GTT |
- AMDGPU_GEM_DOMAIN_CPU |
- AMDGPU_GEM_DOMAIN_GDS |
- AMDGPU_GEM_DOMAIN_GWS |
- AMDGPU_GEM_DOMAIN_OA);
+ bo->prefered_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM |
+ AMDGPU_GEM_DOMAIN_GTT |
+ AMDGPU_GEM_DOMAIN_CPU |
+ AMDGPU_GEM_DOMAIN_GDS |
+ AMDGPU_GEM_DOMAIN_GWS |
+ AMDGPU_GEM_DOMAIN_OA);
+ bo->allowed_domains = bo->prefered_domains;
+ if (!kernel && bo->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
+ bo->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
bo->flags = flags;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index d495db3..8c729b1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -88,8 +88,6 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
struct amdgpu_bo_list_entry *entry)
{
entry->robj = vm->page_directory;
- entry->prefered_domains = AMDGPU_GEM_DOMAIN_VRAM;
- entry->allowed_domains = AMDGPU_GEM_DOMAIN_VRAM;
entry->priority = 0;
entry->tv.bo = &vm->page_directory->tbo;
entry->tv.shared = true;
@@ -1131,8 +1129,6 @@ int amdgpu_vm_bo_map(struct amdgpu_device *adev,
}
entry->robj = pt;
- entry->prefered_domains = AMDGPU_GEM_DOMAIN_VRAM;
- entry->allowed_domains = AMDGPU_GEM_DOMAIN_VRAM;
entry->priority = 0;
entry->tv.bo = &entry->robj->tbo;
entry->tv.shared = true;
--
2.7.4
|
