diff options
author | Antony Clince Alex <aalex@nvidia.com> | 2018-04-25 11:56:21 -0400 |
---|---|---|
committer | mobile promotions <svcmobile_promotions@nvidia.com> | 2018-05-09 21:25:45 -0400 |
commit | 50d1b0c72bfa5bf662c04db85e39c04ca3abfd67 (patch) | |
tree | 85d12588affcfcd23a3794257f965542aadea6b1 /drivers/gpu/nvgpu/common/sim.c | |
parent | b144935644f76328aa9ec2592655b4553ed28222 (diff) |
gpu: nvgpu: os-agnostic segregation of sim/sim_pci
segregated os-agnostic function from linux/sim.c and linux/sim_pci.c
to sim.c and sim_pci.c, while retaining os-specific functions.
renamed all gk20a_* api's to nvgpu_*.
renamed hw_sim_gk20a.h to nvgpu/hw_sim.h
moved hw_sim_pci.h to nvgpu/hw_sim_pci.h
JIRA VQRM-2368
Change-Id: I040a6b12b19111a0b99280245808ea2b0f344cdd
Signed-off-by: Antony Clince Alex <aalex@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1702425
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Diffstat (limited to 'drivers/gpu/nvgpu/common/sim.c')
-rw-r--r-- | drivers/gpu/nvgpu/common/sim.c | 311 |
1 files changed, 311 insertions, 0 deletions
diff --git a/drivers/gpu/nvgpu/common/sim.c b/drivers/gpu/nvgpu/common/sim.c new file mode 100644 index 00000000..0f88ec4d --- /dev/null +++ b/drivers/gpu/nvgpu/common/sim.c | |||
@@ -0,0 +1,311 @@ | |||
1 | /* | ||
2 | * Copyright (c) 2017-2018, NVIDIA CORPORATION. All rights reserved. | ||
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 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER | ||
18 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING | ||
19 | * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER | ||
20 | * DEALINGS IN THE SOFTWARE. | ||
21 | */ | ||
22 | |||
23 | #include <nvgpu/log.h> | ||
24 | #include <nvgpu/bitops.h> | ||
25 | #include <nvgpu/nvgpu_mem.h> | ||
26 | #include <nvgpu/dma.h> | ||
27 | #include <nvgpu/io.h> | ||
28 | #include <nvgpu/hw_sim.h> | ||
29 | #include "gk20a/gk20a.h" | ||
30 | #include "linux/sim.h" | ||
31 | |||
32 | |||
33 | int nvgpu_alloc_sim_buffer(struct gk20a *g, struct nvgpu_mem *mem) | ||
34 | { | ||
35 | int err; | ||
36 | |||
37 | err = nvgpu_dma_alloc(g, PAGE_SIZE, mem); | ||
38 | |||
39 | if (err) | ||
40 | return err; | ||
41 | /* | ||
42 | * create a valid cpu_va mapping | ||
43 | */ | ||
44 | nvgpu_mem_begin(g, mem); | ||
45 | |||
46 | return 0; | ||
47 | } | ||
48 | |||
49 | void nvgpu_free_sim_buffer(struct gk20a *g, struct nvgpu_mem *mem) | ||
50 | { | ||
51 | if (nvgpu_mem_is_valid(mem)) { | ||
52 | /* | ||
53 | * invalidate the cpu_va mapping | ||
54 | */ | ||
55 | nvgpu_mem_end(g, mem); | ||
56 | nvgpu_dma_free(g, mem); | ||
57 | } | ||
58 | |||
59 | memset(mem, 0, sizeof(*mem)); | ||
60 | } | ||
61 | |||
62 | void nvgpu_free_sim_support(struct gk20a *g) | ||
63 | { | ||
64 | nvgpu_free_sim_buffer(g, &g->sim->send_bfr); | ||
65 | nvgpu_free_sim_buffer(g, &g->sim->recv_bfr); | ||
66 | nvgpu_free_sim_buffer(g, &g->sim->msg_bfr); | ||
67 | } | ||
68 | |||
69 | void nvgpu_remove_sim_support(struct gk20a *g) | ||
70 | { | ||
71 | if (g->sim) | ||
72 | nvgpu_free_sim_support(g); | ||
73 | } | ||
74 | |||
75 | static inline u32 sim_msg_header_size(void) | ||
76 | { | ||
77 | return 24;/*TBD: fix the header to gt this from NV_VGPU_MSG_HEADER*/ | ||
78 | } | ||
79 | |||
80 | static inline u32 *sim_msg_bfr(struct gk20a *g, u32 byte_offset) | ||
81 | { | ||
82 | u8 *cpu_va; | ||
83 | |||
84 | cpu_va = (u8 *)g->sim->msg_bfr.cpu_va; | ||
85 | |||
86 | return (u32 *)(cpu_va + byte_offset); | ||
87 | } | ||
88 | |||
89 | static inline u32 *sim_msg_hdr(struct gk20a *g, u32 byte_offset) | ||
90 | { | ||
91 | return sim_msg_bfr(g, byte_offset); /*starts at 0*/ | ||
92 | } | ||
93 | |||
94 | static inline u32 *sim_msg_param(struct gk20a *g, u32 byte_offset) | ||
95 | { | ||
96 | /*starts after msg header/cmn*/ | ||
97 | return sim_msg_bfr(g, byte_offset + sim_msg_header_size()); | ||
98 | } | ||
99 | |||
100 | static inline void sim_write_hdr(struct gk20a *g, u32 func, u32 size) | ||
101 | { | ||
102 | /*memset(g->sim->msg_bfr.kvaddr,0,min(PAGE_SIZE,size));*/ | ||
103 | *sim_msg_hdr(g, sim_msg_signature_r()) = sim_msg_signature_valid_v(); | ||
104 | *sim_msg_hdr(g, sim_msg_result_r()) = sim_msg_result_rpc_pending_v(); | ||
105 | *sim_msg_hdr(g, sim_msg_spare_r()) = sim_msg_spare__init_v(); | ||
106 | *sim_msg_hdr(g, sim_msg_function_r()) = func; | ||
107 | *sim_msg_hdr(g, sim_msg_length_r()) = size + sim_msg_header_size(); | ||
108 | } | ||
109 | |||
110 | static inline u32 sim_escape_read_hdr_size(void) | ||
111 | { | ||
112 | return 12; /*TBD: fix NV_VGPU_SIM_ESCAPE_READ_HEADER*/ | ||
113 | } | ||
114 | |||
115 | static u32 *sim_send_ring_bfr(struct gk20a *g, u32 byte_offset) | ||
116 | { | ||
117 | u8 *cpu_va; | ||
118 | |||
119 | cpu_va = (u8 *)g->sim->send_bfr.cpu_va; | ||
120 | |||
121 | return (u32 *)(cpu_va + byte_offset); | ||
122 | } | ||
123 | |||
124 | static int rpc_send_message(struct gk20a *g) | ||
125 | { | ||
126 | /* calculations done in units of u32s */ | ||
127 | u32 send_base = sim_send_put_pointer_v(g->sim->send_ring_put) * 2; | ||
128 | u32 dma_offset = send_base + sim_dma_r()/sizeof(u32); | ||
129 | u32 dma_hi_offset = send_base + sim_dma_hi_r()/sizeof(u32); | ||
130 | |||
131 | *sim_send_ring_bfr(g, dma_offset*sizeof(u32)) = | ||
132 | sim_dma_target_phys_pci_coherent_f() | | ||
133 | sim_dma_status_valid_f() | | ||
134 | sim_dma_size_4kb_f() | | ||
135 | sim_dma_addr_lo_f(nvgpu_mem_get_addr(g, &g->sim->msg_bfr) | ||
136 | >> PAGE_SHIFT); | ||
137 | |||
138 | *sim_send_ring_bfr(g, dma_hi_offset*sizeof(u32)) = | ||
139 | u64_hi32(nvgpu_mem_get_addr(g, &g->sim->msg_bfr)); | ||
140 | |||
141 | *sim_msg_hdr(g, sim_msg_sequence_r()) = g->sim->sequence_base++; | ||
142 | |||
143 | g->sim->send_ring_put = (g->sim->send_ring_put + 2 * sizeof(u32)) | ||
144 | % PAGE_SIZE; | ||
145 | |||
146 | /* Update the put pointer. This will trap into the host. */ | ||
147 | sim_writel(g->sim, sim_send_put_r(), g->sim->send_ring_put); | ||
148 | |||
149 | return 0; | ||
150 | } | ||
151 | |||
152 | static inline u32 *sim_recv_ring_bfr(struct gk20a *g, u32 byte_offset) | ||
153 | { | ||
154 | u8 *cpu_va; | ||
155 | |||
156 | cpu_va = (u8 *)g->sim->recv_bfr.cpu_va; | ||
157 | |||
158 | return (u32 *)(cpu_va + byte_offset); | ||
159 | } | ||
160 | |||
161 | static int rpc_recv_poll(struct gk20a *g) | ||
162 | { | ||
163 | u64 recv_phys_addr; | ||
164 | |||
165 | /* XXX This read is not required (?) */ | ||
166 | /*pVGpu->recv_ring_get = VGPU_REG_RD32(pGpu, NV_VGPU_RECV_GET);*/ | ||
167 | |||
168 | /* Poll the recv ring get pointer in an infinite loop*/ | ||
169 | do { | ||
170 | g->sim->recv_ring_put = sim_readl(g->sim, sim_recv_put_r()); | ||
171 | } while (g->sim->recv_ring_put == g->sim->recv_ring_get); | ||
172 | |||
173 | /* process all replies */ | ||
174 | while (g->sim->recv_ring_put != g->sim->recv_ring_get) { | ||
175 | /* these are in u32 offsets*/ | ||
176 | u32 dma_lo_offset = | ||
177 | sim_recv_put_pointer_v(g->sim->recv_ring_get)*2 + 0; | ||
178 | u32 dma_hi_offset = dma_lo_offset + 1; | ||
179 | u32 recv_phys_addr_lo = sim_dma_addr_lo_v( | ||
180 | *sim_recv_ring_bfr(g, dma_lo_offset*4)); | ||
181 | u32 recv_phys_addr_hi = sim_dma_hi_addr_v( | ||
182 | *sim_recv_ring_bfr(g, dma_hi_offset*4)); | ||
183 | |||
184 | recv_phys_addr = (u64)recv_phys_addr_hi << 32 | | ||
185 | (u64)recv_phys_addr_lo << PAGE_SHIFT; | ||
186 | |||
187 | if (recv_phys_addr != | ||
188 | nvgpu_mem_get_addr(g, &g->sim->msg_bfr)) { | ||
189 | nvgpu_err(g, "%s Error in RPC reply", | ||
190 | __func__); | ||
191 | return -1; | ||
192 | } | ||
193 | |||
194 | /* Update GET pointer */ | ||
195 | g->sim->recv_ring_get = (g->sim->recv_ring_get + 2*sizeof(u32)) | ||
196 | % PAGE_SIZE; | ||
197 | |||
198 | sim_writel(g->sim, sim_recv_get_r(), g->sim->recv_ring_get); | ||
199 | |||
200 | g->sim->recv_ring_put = sim_readl(g->sim, sim_recv_put_r()); | ||
201 | } | ||
202 | |||
203 | return 0; | ||
204 | } | ||
205 | |||
206 | static int issue_rpc_and_wait(struct gk20a *g) | ||
207 | { | ||
208 | int err; | ||
209 | |||
210 | err = rpc_send_message(g); | ||
211 | if (err) { | ||
212 | nvgpu_err(g, "%s failed rpc_send_message", | ||
213 | __func__); | ||
214 | return err; | ||
215 | } | ||
216 | |||
217 | err = rpc_recv_poll(g); | ||
218 | if (err) { | ||
219 | nvgpu_err(g, "%s failed rpc_recv_poll", | ||
220 | __func__); | ||
221 | return err; | ||
222 | } | ||
223 | |||
224 | /* Now check if RPC really succeeded */ | ||
225 | if (*sim_msg_hdr(g, sim_msg_result_r()) != sim_msg_result_success_v()) { | ||
226 | nvgpu_err(g, "%s received failed status!", | ||
227 | __func__); | ||
228 | return -(*sim_msg_hdr(g, sim_msg_result_r())); | ||
229 | } | ||
230 | return 0; | ||
231 | } | ||
232 | |||
233 | static int nvgpu_sim_esc_readl(struct gk20a *g, | ||
234 | char *path, u32 index, u32 *data) | ||
235 | { | ||
236 | int err; | ||
237 | size_t pathlen = strlen(path); | ||
238 | u32 data_offset; | ||
239 | |||
240 | sim_write_hdr(g, sim_msg_function_sim_escape_read_v(), | ||
241 | sim_escape_read_hdr_size()); | ||
242 | *sim_msg_param(g, 0) = index; | ||
243 | *sim_msg_param(g, 4) = sizeof(u32); | ||
244 | data_offset = roundup(0xc + pathlen + 1, sizeof(u32)); | ||
245 | *sim_msg_param(g, 8) = data_offset; | ||
246 | strcpy((char *)sim_msg_param(g, 0xc), path); | ||
247 | |||
248 | err = issue_rpc_and_wait(g); | ||
249 | |||
250 | if (!err) | ||
251 | memcpy(data, sim_msg_param(g, data_offset), sizeof(u32)); | ||
252 | return err; | ||
253 | } | ||
254 | |||
255 | int nvgpu_init_sim_support(struct gk20a *g) | ||
256 | { | ||
257 | int err = -ENOMEM; | ||
258 | u64 phys; | ||
259 | |||
260 | if (!g->sim) | ||
261 | return 0; | ||
262 | |||
263 | /* allocate sim event/msg buffers */ | ||
264 | err = nvgpu_alloc_sim_buffer(g, &g->sim->send_bfr); | ||
265 | err = err || nvgpu_alloc_sim_buffer(g, &g->sim->recv_bfr); | ||
266 | err = err || nvgpu_alloc_sim_buffer(g, &g->sim->msg_bfr); | ||
267 | |||
268 | if (err) | ||
269 | goto fail; | ||
270 | /*mark send ring invalid*/ | ||
271 | sim_writel(g->sim, sim_send_ring_r(), sim_send_ring_status_invalid_f()); | ||
272 | |||
273 | /*read get pointer and make equal to put*/ | ||
274 | g->sim->send_ring_put = sim_readl(g->sim, sim_send_get_r()); | ||
275 | sim_writel(g->sim, sim_send_put_r(), g->sim->send_ring_put); | ||
276 | |||
277 | /*write send ring address and make it valid*/ | ||
278 | phys = nvgpu_mem_get_addr(g, &g->sim->send_bfr); | ||
279 | sim_writel(g->sim, sim_send_ring_hi_r(), | ||
280 | sim_send_ring_hi_addr_f(u64_hi32(phys))); | ||
281 | sim_writel(g->sim, sim_send_ring_r(), | ||
282 | sim_send_ring_status_valid_f() | | ||
283 | sim_send_ring_target_phys_pci_coherent_f() | | ||
284 | sim_send_ring_size_4kb_f() | | ||
285 | sim_send_ring_addr_lo_f(phys >> PAGE_SHIFT)); | ||
286 | |||
287 | /*repeat for recv ring (but swap put,get as roles are opposite) */ | ||
288 | sim_writel(g->sim, sim_recv_ring_r(), sim_recv_ring_status_invalid_f()); | ||
289 | |||
290 | /*read put pointer and make equal to get*/ | ||
291 | g->sim->recv_ring_get = sim_readl(g->sim, sim_recv_put_r()); | ||
292 | sim_writel(g->sim, sim_recv_get_r(), g->sim->recv_ring_get); | ||
293 | |||
294 | /*write send ring address and make it valid*/ | ||
295 | phys = nvgpu_mem_get_addr(g, &g->sim->recv_bfr); | ||
296 | sim_writel(g->sim, sim_recv_ring_hi_r(), | ||
297 | sim_recv_ring_hi_addr_f(u64_hi32(phys))); | ||
298 | sim_writel(g->sim, sim_recv_ring_r(), | ||
299 | sim_recv_ring_status_valid_f() | | ||
300 | sim_recv_ring_target_phys_pci_coherent_f() | | ||
301 | sim_recv_ring_size_4kb_f() | | ||
302 | sim_recv_ring_addr_lo_f(phys >> PAGE_SHIFT)); | ||
303 | |||
304 | g->sim->remove_support = nvgpu_remove_sim_support; | ||
305 | g->sim->esc_readl = nvgpu_sim_esc_readl; | ||
306 | return 0; | ||
307 | |||
308 | fail: | ||
309 | nvgpu_free_sim_support(g); | ||
310 | return err; | ||
311 | } | ||