Rev 1404 | Rev 1430 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
1120 | serge | 1 | /* |
2 | * Copyright 2008 Advanced Micro Devices, Inc. |
||
3 | * Copyright 2008 Red Hat Inc. |
||
4 | * Copyright 2009 Jerome Glisse. |
||
5 | * |
||
6 | * Permission is hereby granted, free of charge, to any person obtaining a |
||
7 | * copy of this software and associated documentation files (the "Software"), |
||
8 | * to deal in the Software without restriction, including without limitation |
||
9 | * the rights to use, copy, modify, merge, publish, distribute, sublicense, |
||
10 | * and/or sell copies of the Software, and to permit persons to whom the |
||
11 | * Software is furnished to do so, subject to the following conditions: |
||
12 | * |
||
13 | * The above copyright notice and this permission notice shall be included in |
||
14 | * all copies or substantial portions of the Software. |
||
15 | * |
||
16 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
||
17 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
||
18 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL |
||
19 | * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR |
||
20 | * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, |
||
21 | * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR |
||
22 | * OTHER DEALINGS IN THE SOFTWARE. |
||
23 | * |
||
24 | * Authors: Dave Airlie |
||
25 | * Alex Deucher |
||
26 | * Jerome Glisse |
||
27 | */ |
||
1125 | serge | 28 | #include "drmP.h" |
1120 | serge | 29 | #include "radeon_drm.h" |
30 | #include "radeon.h" |
||
31 | #include "radeon_reg.h" |
||
32 | |||
33 | /* |
||
34 | * Common GART table functions. |
||
35 | */ |
||
36 | int radeon_gart_table_ram_alloc(struct radeon_device *rdev) |
||
37 | { |
||
38 | void *ptr; |
||
39 | |||
1246 | serge | 40 | ptr = pci_alloc_consistent(rdev->pdev, rdev->gart.table_size, |
41 | &rdev->gart.table_addr); |
||
1120 | serge | 42 | if (ptr == NULL) { |
43 | return -ENOMEM; |
||
44 | } |
||
45 | #ifdef CONFIG_X86 |
||
46 | if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480 || |
||
47 | rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) { |
||
48 | set_memory_uc((unsigned long)ptr, |
||
49 | rdev->gart.table_size >> PAGE_SHIFT); |
||
50 | } |
||
51 | #endif |
||
52 | rdev->gart.table.ram.ptr = ptr; |
||
53 | memset((void *)rdev->gart.table.ram.ptr, 0, rdev->gart.table_size); |
||
54 | return 0; |
||
55 | } |
||
56 | |||
57 | void radeon_gart_table_ram_free(struct radeon_device *rdev) |
||
58 | { |
||
59 | if (rdev->gart.table.ram.ptr == NULL) { |
||
60 | return; |
||
61 | } |
||
62 | #ifdef CONFIG_X86 |
||
63 | if (rdev->family == CHIP_RS400 || rdev->family == CHIP_RS480 || |
||
64 | rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740) { |
||
65 | set_memory_wb((unsigned long)rdev->gart.table.ram.ptr, |
||
66 | rdev->gart.table_size >> PAGE_SHIFT); |
||
67 | } |
||
68 | #endif |
||
1128 | serge | 69 | // pci_free_consistent(rdev->pdev, rdev->gart.table_size, |
70 | // (void *)rdev->gart.table.ram.ptr, |
||
71 | // rdev->gart.table_addr); |
||
1120 | serge | 72 | rdev->gart.table.ram.ptr = NULL; |
73 | rdev->gart.table_addr = 0; |
||
74 | } |
||
75 | |||
76 | int radeon_gart_table_vram_alloc(struct radeon_device *rdev) |
||
77 | { |
||
78 | int r; |
||
79 | |||
80 | if (rdev->gart.table.vram.robj == NULL) { |
||
1404 | serge | 81 | r = radeon_bo_create(rdev, NULL, rdev->gart.table_size, |
82 | true, RADEON_GEM_DOMAIN_VRAM, |
||
83 | &rdev->gart.table.vram.robj); |
||
1120 | serge | 84 | if (r) { |
85 | return r; |
||
86 | } |
||
87 | } |
||
1179 | serge | 88 | return 0; |
89 | } |
||
90 | |||
91 | int radeon_gart_table_vram_pin(struct radeon_device *rdev) |
||
92 | { |
||
93 | uint64_t gpu_addr; |
||
94 | int r; |
||
95 | |||
1404 | serge | 96 | r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); |
97 | if (unlikely(r != 0)) |
||
98 | return r; |
||
99 | r = radeon_bo_pin(rdev->gart.table.vram.robj, |
||
1120 | serge | 100 | RADEON_GEM_DOMAIN_VRAM, &gpu_addr); |
101 | if (r) { |
||
1404 | serge | 102 | radeon_bo_unreserve(rdev->gart.table.vram.robj); |
1120 | serge | 103 | return r; |
104 | } |
||
1404 | serge | 105 | r = radeon_bo_kmap(rdev->gart.table.vram.robj, |
1120 | serge | 106 | (void **)&rdev->gart.table.vram.ptr); |
1404 | serge | 107 | if (r) |
108 | radeon_bo_unpin(rdev->gart.table.vram.robj); |
||
109 | radeon_bo_unreserve(rdev->gart.table.vram.robj); |
||
110 | rdev->gart.table_addr = gpu_addr; |
||
111 | return r; |
||
1120 | serge | 112 | } |
113 | |||
114 | void radeon_gart_table_vram_free(struct radeon_device *rdev) |
||
115 | { |
||
1404 | serge | 116 | int r; |
117 | |||
1120 | serge | 118 | if (rdev->gart.table.vram.robj == NULL) { |
119 | return; |
||
120 | } |
||
1404 | serge | 121 | r = radeon_bo_reserve(rdev->gart.table.vram.robj, false); |
122 | if (likely(r == 0)) { |
||
123 | radeon_bo_kunmap(rdev->gart.table.vram.robj); |
||
124 | radeon_bo_unpin(rdev->gart.table.vram.robj); |
||
125 | radeon_bo_unreserve(rdev->gart.table.vram.robj); |
||
126 | } |
||
127 | radeon_bo_unref(&rdev->gart.table.vram.robj); |
||
1120 | serge | 128 | } |
129 | |||
130 | |||
131 | |||
132 | |||
133 | /* |
||
134 | * Common gart functions. |
||
135 | */ |
||
136 | void radeon_gart_unbind(struct radeon_device *rdev, unsigned offset, |
||
137 | int pages) |
||
138 | { |
||
139 | unsigned t; |
||
140 | unsigned p; |
||
141 | int i, j; |
||
142 | |||
143 | if (!rdev->gart.ready) { |
||
1404 | serge | 144 | WARN(1, "trying to unbind memory to unitialized GART !\n"); |
1120 | serge | 145 | return; |
146 | } |
||
1268 | serge | 147 | t = offset / RADEON_GPU_PAGE_SIZE; |
148 | p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); |
||
1120 | serge | 149 | for (i = 0; i < pages; i++, p++) { |
150 | if (rdev->gart.pages[p]) { |
||
151 | // pci_unmap_page(rdev->pdev, rdev->gart.pages_addr[p], |
||
152 | // PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); |
||
153 | rdev->gart.pages[p] = NULL; |
||
154 | rdev->gart.pages_addr[p] = 0; |
||
1268 | serge | 155 | for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { |
1120 | serge | 156 | radeon_gart_set_page(rdev, t, 0); |
157 | } |
||
158 | } |
||
159 | } |
||
160 | mb(); |
||
161 | radeon_gart_tlb_flush(rdev); |
||
162 | } |
||
163 | |||
164 | int radeon_gart_bind(struct radeon_device *rdev, unsigned offset, |
||
165 | int pages, u32_t *pagelist) |
||
166 | { |
||
167 | unsigned t; |
||
168 | unsigned p; |
||
169 | uint64_t page_base; |
||
170 | int i, j; |
||
171 | |||
1179 | serge | 172 | ENTER(); |
173 | |||
1120 | serge | 174 | dbgprintf("offset %x pages %x list %x\n", |
175 | offset, pages, pagelist); |
||
176 | |||
177 | if (!rdev->gart.ready) { |
||
178 | DRM_ERROR("trying to bind memory to unitialized GART !\n"); |
||
179 | return -EINVAL; |
||
180 | } |
||
1268 | serge | 181 | t = offset / RADEON_GPU_PAGE_SIZE; |
182 | p = t / (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); |
||
1120 | serge | 183 | |
184 | for (i = 0; i < pages; i++, p++) { |
||
185 | /* we need to support large memory configurations */ |
||
186 | /* assume that unbind have already been call on the range */ |
||
187 | |||
188 | rdev->gart.pages_addr[p] = pagelist[i] & ~4095; |
||
189 | |||
190 | //if (pci_dma_mapping_error(rdev->pdev, rdev->gart.pages_addr[p])) { |
||
191 | // /* FIXME: failed to map page (return -ENOMEM?) */ |
||
192 | // radeon_gart_unbind(rdev, offset, pages); |
||
193 | // return -ENOMEM; |
||
194 | //} |
||
195 | rdev->gart.pages[p] = pagelist[i]; |
||
1268 | serge | 196 | page_base = rdev->gart.pages_addr[p]; |
197 | for (j = 0; j < (PAGE_SIZE / RADEON_GPU_PAGE_SIZE); j++, t++) { |
||
1120 | serge | 198 | radeon_gart_set_page(rdev, t, page_base); |
1268 | serge | 199 | page_base += RADEON_GPU_PAGE_SIZE; |
1120 | serge | 200 | } |
201 | } |
||
202 | mb(); |
||
203 | radeon_gart_tlb_flush(rdev); |
||
204 | return 0; |
||
205 | } |
||
206 | |||
207 | int radeon_gart_init(struct radeon_device *rdev) |
||
208 | { |
||
209 | if (rdev->gart.pages) { |
||
210 | return 0; |
||
211 | } |
||
1268 | serge | 212 | /* We need PAGE_SIZE >= RADEON_GPU_PAGE_SIZE */ |
213 | if (PAGE_SIZE < RADEON_GPU_PAGE_SIZE) { |
||
1120 | serge | 214 | DRM_ERROR("Page size is smaller than GPU page size!\n"); |
215 | return -EINVAL; |
||
216 | } |
||
217 | /* Compute table size */ |
||
218 | rdev->gart.num_cpu_pages = rdev->mc.gtt_size / PAGE_SIZE; |
||
1268 | serge | 219 | rdev->gart.num_gpu_pages = rdev->mc.gtt_size / RADEON_GPU_PAGE_SIZE; |
1120 | serge | 220 | DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n", |
221 | rdev->gart.num_cpu_pages, rdev->gart.num_gpu_pages); |
||
222 | /* Allocate pages table */ |
||
223 | rdev->gart.pages = kzalloc(sizeof(void *) * rdev->gart.num_cpu_pages, |
||
224 | GFP_KERNEL); |
||
225 | if (rdev->gart.pages == NULL) { |
||
1404 | serge | 226 | radeon_gart_fini(rdev); |
1120 | serge | 227 | return -ENOMEM; |
228 | } |
||
1404 | serge | 229 | rdev->gart.pages_addr = kzalloc(sizeof(dma_addr_t) * |
1120 | serge | 230 | rdev->gart.num_cpu_pages, GFP_KERNEL); |
231 | if (rdev->gart.pages_addr == NULL) { |
||
1404 | serge | 232 | radeon_gart_fini(rdev); |
1120 | serge | 233 | return -ENOMEM; |
234 | } |
||
235 | return 0; |
||
236 | } |
||
237 | |||
238 | void radeon_gart_fini(struct radeon_device *rdev) |
||
239 | { |
||
240 | if (rdev->gart.pages && rdev->gart.pages_addr && rdev->gart.ready) { |
||
241 | /* unbind pages */ |
||
242 | radeon_gart_unbind(rdev, 0, rdev->gart.num_cpu_pages); |
||
243 | } |
||
244 | rdev->gart.ready = false; |
||
245 | kfree(rdev->gart.pages); |
||
246 | kfree(rdev->gart.pages_addr); |
||
247 | rdev->gart.pages = NULL; |
||
248 | rdev->gart.pages_addr = NULL; |
||
249 | }>>>>> |