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
|
/*
* Copyright 2019 Advanced Micro Devices, Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Tom St Denis <tom.stdenis@amd.com>
*
*/
#include "umr.h"
#include <inttypes.h>
/**
* umr_sdma_decode_ring - Read a GPU ring and decode into a sdma stream
*
* @ringname - Common name of the ring, e.g., 'gfx' or 'comp_1.0.0'
* @no_halt - Set to 0 to issue an SQ_CMD halt command
*
* Return a sdma stream if successful.
*/
struct umr_sdma_stream *umr_sdma_decode_ring(struct umr_asic *asic, char *ringname)
{
void *ps;
uint32_t *ringdata, ringsize;
// read ring data and reduce indeices modulo ring size
// since the kernel returned values might be unwrapped.
ringdata = umr_read_ring_data(asic, ringname, &ringsize);
ringsize /= 4;
ringdata[0] %= ringsize;
ringdata[1] %= ringsize;
// only proceed if there is data to read
// and then linearize it so that the stream
// decoder can do it's thing
if (ringdata[0] != ringdata[1]) { // rptr != wptr
uint32_t *lineardata, linearsize;
// copy ring data into linear array
lineardata = calloc(ringsize, sizeof(*lineardata));
linearsize = 0;
while (ringdata[0] != ringdata[1]) {
lineardata[linearsize++] = ringdata[3 + ringdata[0]]; // first 3 words are rptr/wptr/dwptr
ringdata[0] = (ringdata[0] + 1) % ringsize;
}
ps = umr_sdma_decode_stream(asic, 0, lineardata, linearsize);
free(lineardata);
free(ringdata);
} else {
ps = NULL;
}
return ps;
}
/**
* umr_sdma_decode_stream - Decode an array of sdma packets into a sdma stream
*
* @vmid: The VMID (or zero) that this array comes from (if say an IB)
* @stream: An array of DWORDS which contain the sdma packets
* @nwords: The number of words in the stream
*
* Returns a sdma stream if successfully decoded.
*/
struct umr_sdma_stream *umr_sdma_decode_stream(struct umr_asic *asic, int vmid, uint32_t *stream, uint32_t nwords)
{
struct umr_sdma_stream *ops, *ps;
ps = ops = calloc(1, sizeof *ops);
if (!ps) {
fprintf(stderr, "[ERROR]: Out of memory\n");
return NULL;
}
while (nwords) {
ps->opcode = *stream & 0xFF;
ps->sub_opcode = (*stream >> 8) & 0xFF;
ps->header_dw = *stream++;
switch (ps->opcode) {
case 0: // NOP
ps->nwords = 0; // no words other than header
break;
case 1: // COPY
switch (ps->sub_opcode) {
case 0: // LINEAR
ps->nwords = 6;
// BROADCAST
if (ps->header_dw & (1UL << 27)) {
ps->nwords += 2;
}
break;
case 1: // TILED
ps->nwords = 11;
break;
case 3: // STRUCTURE/SOA
ps->nwords = 7;
break;
case 4: // LINEAR_SUB_WINDOW
ps->nwords = 12;
break;
case 5: // TILED_SUB_WINDOW
ps->nwords = 13;
break;
case 6: // T2T_SUB_WIND
ps->nwords = 14;
break;
}
break;
case 2: // WRITE
switch (ps->sub_opcode) {
case 0: // LINEAR
ps->nwords = 4;
ps->nwords += stream[2] - 1;
break;
case 1: // TILED
ps->nwords = 9;
break;
}
break;
case 4: // INDIRECT
ps->ib.vmid = (ps->header_dw >> 16) & 0xF;
ps->ib.addr = ((uint64_t)stream[1] << 32) | stream[0];
ps->ib.size = stream[2];
if (asic->family >= FAMILY_AI)
ps->ib.vmid |= UMR_MM_HUB;
ps->nwords = 5;
{
uint32_t *data = calloc(sizeof(*data), ps->ib.size);
if (umr_read_vram(asic, ps->ib.vmid, ps->ib.addr, ps->ib.size * sizeof(*data), data) == 0)
ps->next_ib = umr_sdma_decode_stream(asic, ps->ib.vmid, data, ps->ib.size);
free(data);
}
break;
case 5: // FENCE
ps->nwords = 3;
break;
case 6: // TRAP
ps->nwords = 1;
break;
case 7: // SEM
ps->nwords = 2;
break;
case 8: // POLL_REGMEM
ps->nwords = ps->sub_opcode ? 3 : 5;
break;
case 9: // COND_EXE
ps->nwords = 4;
break;
case 10: // ATOMIC
ps->nwords = 7;
break;
case 11: // CONST_FILL
ps->nwords = 4;
break;
case 12: // GEN_PTEPDE
ps->nwords = 9;
break;
case 13: // TIMESTAMP
switch (ps->sub_opcode) {
case 0:
ps->nwords = 2;
break;
case 1:
ps->nwords = 2;
break;
case 2:
ps->nwords = 2;
break;
}
break;
case 14: // SRBM_WRITE
ps->nwords = 2;
break;
case 15: // PRE_EXE
ps->nwords = 1;
break;
}
ps->words = calloc(ps->nwords, sizeof(ps->words[0]));
memcpy(ps->words, stream, ps->nwords * sizeof(ps->words[0]));
stream += ps->nwords;
if (nwords <= 1 + ps->nwords) {
fprintf(stderr, "[WARNING]: Ran out of stream words in SDMA stream decode\n");
return ops;
}
nwords -= 1 + ps->nwords;
ps->next = calloc(1, sizeof(ps->next[0]));
ps = ps->next;
}
return ops;
}
/**
* umr_free_sdma_stream - Free a sdma stream object
*/
void umr_free_sdma_stream(struct umr_sdma_stream *stream)
{
while (stream) {
struct umr_sdma_stream *n;
n = stream->next;
if (stream->next_ib)
umr_free_sdma_stream(stream->next_ib);
free(stream->words);
free(stream);
stream = n;
}
}
|
