Michel Dänzer | ecc0b32 | 2009-07-21 11:23:57 +0200 | [diff] [blame] | 1 | /* |
| 2 | * Copyright 2009 VMware, 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 | * Authors: Michel Dänzer |
| 23 | */ |
| 24 | #include <drm/drmP.h> |
| 25 | #include <drm/radeon_drm.h> |
| 26 | #include "radeon_reg.h" |
| 27 | #include "radeon.h" |
| 28 | |
| 29 | |
| 30 | /* Test BO GTT->VRAM and VRAM->GTT GPU copies across the whole GTT aperture */ |
| 31 | void radeon_test_moves(struct radeon_device *rdev) |
| 32 | { |
| 33 | struct radeon_object *vram_obj = NULL; |
| 34 | struct radeon_object **gtt_obj = NULL; |
| 35 | struct radeon_fence *fence = NULL; |
| 36 | uint64_t gtt_addr, vram_addr; |
| 37 | unsigned i, n, size; |
| 38 | int r; |
| 39 | |
| 40 | size = 1024 * 1024; |
| 41 | |
| 42 | /* Number of tests = |
| 43 | * (Total GTT - IB pool - writeback page - ring buffer) / test size |
| 44 | */ |
| 45 | n = (rdev->mc.gtt_size - RADEON_IB_POOL_SIZE*64*1024 - 4096 - |
| 46 | rdev->cp.ring_size) / size; |
| 47 | |
| 48 | gtt_obj = kzalloc(n * sizeof(*gtt_obj), GFP_KERNEL); |
| 49 | if (!gtt_obj) { |
| 50 | DRM_ERROR("Failed to allocate %d pointers\n", n); |
| 51 | r = 1; |
| 52 | goto out_cleanup; |
| 53 | } |
| 54 | |
| 55 | r = radeon_object_create(rdev, NULL, size, true, RADEON_GEM_DOMAIN_VRAM, |
| 56 | false, &vram_obj); |
| 57 | if (r) { |
| 58 | DRM_ERROR("Failed to create VRAM object\n"); |
| 59 | goto out_cleanup; |
| 60 | } |
| 61 | |
| 62 | r = radeon_object_pin(vram_obj, RADEON_GEM_DOMAIN_VRAM, &vram_addr); |
| 63 | if (r) { |
| 64 | DRM_ERROR("Failed to pin VRAM object\n"); |
| 65 | goto out_cleanup; |
| 66 | } |
| 67 | |
| 68 | for (i = 0; i < n; i++) { |
| 69 | void *gtt_map, *vram_map; |
| 70 | void **gtt_start, **gtt_end; |
| 71 | void **vram_start, **vram_end; |
| 72 | |
| 73 | r = radeon_object_create(rdev, NULL, size, true, |
| 74 | RADEON_GEM_DOMAIN_GTT, false, gtt_obj + i); |
| 75 | if (r) { |
| 76 | DRM_ERROR("Failed to create GTT object %d\n", i); |
| 77 | goto out_cleanup; |
| 78 | } |
| 79 | |
| 80 | r = radeon_object_pin(gtt_obj[i], RADEON_GEM_DOMAIN_GTT, >t_addr); |
| 81 | if (r) { |
| 82 | DRM_ERROR("Failed to pin GTT object %d\n", i); |
| 83 | goto out_cleanup; |
| 84 | } |
| 85 | |
| 86 | r = radeon_object_kmap(gtt_obj[i], >t_map); |
| 87 | if (r) { |
| 88 | DRM_ERROR("Failed to map GTT object %d\n", i); |
| 89 | goto out_cleanup; |
| 90 | } |
| 91 | |
| 92 | for (gtt_start = gtt_map, gtt_end = gtt_map + size; |
| 93 | gtt_start < gtt_end; |
| 94 | gtt_start++) |
| 95 | *gtt_start = gtt_start; |
| 96 | |
| 97 | radeon_object_kunmap(gtt_obj[i]); |
| 98 | |
| 99 | r = radeon_fence_create(rdev, &fence); |
| 100 | if (r) { |
| 101 | DRM_ERROR("Failed to create GTT->VRAM fence %d\n", i); |
| 102 | goto out_cleanup; |
| 103 | } |
| 104 | |
| 105 | r = radeon_copy(rdev, gtt_addr, vram_addr, size / 4096, fence); |
| 106 | if (r) { |
| 107 | DRM_ERROR("Failed GTT->VRAM copy %d\n", i); |
| 108 | goto out_cleanup; |
| 109 | } |
| 110 | |
| 111 | r = radeon_fence_wait(fence, false); |
| 112 | if (r) { |
| 113 | DRM_ERROR("Failed to wait for GTT->VRAM fence %d\n", i); |
| 114 | goto out_cleanup; |
| 115 | } |
| 116 | |
| 117 | radeon_fence_unref(&fence); |
| 118 | |
| 119 | r = radeon_object_kmap(vram_obj, &vram_map); |
| 120 | if (r) { |
| 121 | DRM_ERROR("Failed to map VRAM object after copy %d\n", i); |
| 122 | goto out_cleanup; |
| 123 | } |
| 124 | |
| 125 | for (gtt_start = gtt_map, gtt_end = gtt_map + size, |
| 126 | vram_start = vram_map, vram_end = vram_map + size; |
| 127 | vram_start < vram_end; |
| 128 | gtt_start++, vram_start++) { |
| 129 | if (*vram_start != gtt_start) { |
| 130 | DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, " |
| 131 | "expected 0x%p (GTT map 0x%p-0x%p)\n", |
| 132 | i, *vram_start, gtt_start, gtt_map, |
| 133 | gtt_end); |
| 134 | radeon_object_kunmap(vram_obj); |
| 135 | goto out_cleanup; |
| 136 | } |
| 137 | *vram_start = vram_start; |
| 138 | } |
| 139 | |
| 140 | radeon_object_kunmap(vram_obj); |
| 141 | |
| 142 | r = radeon_fence_create(rdev, &fence); |
| 143 | if (r) { |
| 144 | DRM_ERROR("Failed to create VRAM->GTT fence %d\n", i); |
| 145 | goto out_cleanup; |
| 146 | } |
| 147 | |
| 148 | r = radeon_copy(rdev, vram_addr, gtt_addr, size / 4096, fence); |
| 149 | if (r) { |
| 150 | DRM_ERROR("Failed VRAM->GTT copy %d\n", i); |
| 151 | goto out_cleanup; |
| 152 | } |
| 153 | |
| 154 | r = radeon_fence_wait(fence, false); |
| 155 | if (r) { |
| 156 | DRM_ERROR("Failed to wait for VRAM->GTT fence %d\n", i); |
| 157 | goto out_cleanup; |
| 158 | } |
| 159 | |
| 160 | radeon_fence_unref(&fence); |
| 161 | |
| 162 | r = radeon_object_kmap(gtt_obj[i], >t_map); |
| 163 | if (r) { |
| 164 | DRM_ERROR("Failed to map GTT object after copy %d\n", i); |
| 165 | goto out_cleanup; |
| 166 | } |
| 167 | |
| 168 | for (gtt_start = gtt_map, gtt_end = gtt_map + size, |
| 169 | vram_start = vram_map, vram_end = vram_map + size; |
| 170 | gtt_start < gtt_end; |
| 171 | gtt_start++, vram_start++) { |
| 172 | if (*gtt_start != vram_start) { |
| 173 | DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, " |
| 174 | "expected 0x%p (VRAM map 0x%p-0x%p)\n", |
| 175 | i, *gtt_start, vram_start, vram_map, |
| 176 | vram_end); |
| 177 | radeon_object_kunmap(gtt_obj[i]); |
| 178 | goto out_cleanup; |
| 179 | } |
| 180 | } |
| 181 | |
| 182 | radeon_object_kunmap(gtt_obj[i]); |
| 183 | |
| 184 | DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n", |
| 185 | gtt_addr - rdev->mc.gtt_location); |
| 186 | } |
| 187 | |
| 188 | out_cleanup: |
| 189 | if (vram_obj) { |
| 190 | radeon_object_unpin(vram_obj); |
| 191 | radeon_object_unref(&vram_obj); |
| 192 | } |
| 193 | if (gtt_obj) { |
| 194 | for (i = 0; i < n; i++) { |
| 195 | if (gtt_obj[i]) { |
| 196 | radeon_object_unpin(gtt_obj[i]); |
| 197 | radeon_object_unref(>t_obj[i]); |
| 198 | } |
| 199 | } |
| 200 | kfree(gtt_obj); |
| 201 | } |
| 202 | if (fence) { |
| 203 | radeon_fence_unref(&fence); |
| 204 | } |
| 205 | if (r) { |
| 206 | printk(KERN_WARNING "Error while testing BO move.\n"); |
| 207 | } |
| 208 | } |
| 209 | |