diff options
author | Jerome Glisse <jglisse@redhat.com> | 2009-06-05 08:42:42 -0400 |
---|---|---|
committer | Dave Airlie <airlied@redhat.com> | 2009-06-14 22:01:53 -0400 |
commit | 771fe6b912fca54f03e8a72eb63058b582775362 (patch) | |
tree | 58aa5469ba8058c2b564d50807395ad6cd7bd7e4 /drivers/gpu/drm/radeon/r300.c | |
parent | ba4e7d973dd09b66912ac4c0856add8b0703a997 (diff) |
drm/radeon: introduce kernel modesetting for radeon hardware
Add kernel modesetting support to radeon driver, use the ttm memory
manager to manage memory and DRM/GEM to provide userspace API.
In order to avoid backward compatibility issue and to allow clean
design and code the radeon kernel modesetting use different code path
than old radeon/drm driver.
When kernel modesetting is enabled the IOCTL of radeon/drm
driver are considered as invalid and an error message is printed
in the log and they return failure.
KMS enabled userspace will use new API to talk with the radeon/drm
driver. The new API provide functions to create/destroy/share/mmap
buffer object which are then managed by the kernel memory manager
(here TTM). In order to submit command to the GPU the userspace
provide a buffer holding the command stream, along this buffer
userspace have to provide a list of buffer object used by the
command stream. The kernel radeon driver will then place buffer
in GPU accessible memory and will update command stream to reflect
the position of the different buffers.
The kernel will also perform security check on command stream
provided by the user, we want to catch and forbid any illegal use
of the GPU such as DMA into random system memory or into memory
not owned by the process supplying the command stream. This part
of the code is still incomplete and this why we propose that patch
as a staging driver addition, future security might forbid current
experimental userspace to run.
This code support the following hardware : R1XX,R2XX,R3XX,R4XX,R5XX
(radeon up to X1950). Works is underway to provide support for R6XX,
R7XX and newer hardware (radeon from HD2XXX to HD4XXX).
Authors:
Jerome Glisse <jglisse@redhat.com>
Dave Airlie <airlied@redhat.com>
Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Diffstat (limited to 'drivers/gpu/drm/radeon/r300.c')
-rw-r--r-- | drivers/gpu/drm/radeon/r300.c | 1116 |
1 files changed, 1116 insertions, 0 deletions
diff --git a/drivers/gpu/drm/radeon/r300.c b/drivers/gpu/drm/radeon/r300.c new file mode 100644 index 000000000000..f5870a099d4f --- /dev/null +++ b/drivers/gpu/drm/radeon/r300.c | |||
@@ -0,0 +1,1116 @@ | |||
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 | */ | ||
28 | #include <linux/seq_file.h> | ||
29 | #include "drmP.h" | ||
30 | #include "drm.h" | ||
31 | #include "radeon_reg.h" | ||
32 | #include "radeon.h" | ||
33 | |||
34 | /* r300,r350,rv350,rv370,rv380 depends on : */ | ||
35 | void r100_hdp_reset(struct radeon_device *rdev); | ||
36 | int r100_cp_reset(struct radeon_device *rdev); | ||
37 | int r100_rb2d_reset(struct radeon_device *rdev); | ||
38 | int r100_cp_init(struct radeon_device *rdev, unsigned ring_size); | ||
39 | int r100_pci_gart_enable(struct radeon_device *rdev); | ||
40 | void r100_pci_gart_disable(struct radeon_device *rdev); | ||
41 | void r100_mc_setup(struct radeon_device *rdev); | ||
42 | void r100_mc_disable_clients(struct radeon_device *rdev); | ||
43 | int r100_gui_wait_for_idle(struct radeon_device *rdev); | ||
44 | int r100_cs_packet_parse(struct radeon_cs_parser *p, | ||
45 | struct radeon_cs_packet *pkt, | ||
46 | unsigned idx); | ||
47 | int r100_cs_packet_next_reloc(struct radeon_cs_parser *p, | ||
48 | struct radeon_cs_reloc **cs_reloc); | ||
49 | int r100_cs_parse_packet0(struct radeon_cs_parser *p, | ||
50 | struct radeon_cs_packet *pkt, | ||
51 | unsigned *auth, unsigned n, | ||
52 | radeon_packet0_check_t check); | ||
53 | int r100_cs_parse_packet3(struct radeon_cs_parser *p, | ||
54 | struct radeon_cs_packet *pkt, | ||
55 | unsigned *auth, unsigned n, | ||
56 | radeon_packet3_check_t check); | ||
57 | void r100_cs_dump_packet(struct radeon_cs_parser *p, | ||
58 | struct radeon_cs_packet *pkt); | ||
59 | |||
60 | /* This files gather functions specifics to: | ||
61 | * r300,r350,rv350,rv370,rv380 | ||
62 | * | ||
63 | * Some of these functions might be used by newer ASICs. | ||
64 | */ | ||
65 | void r300_gpu_init(struct radeon_device *rdev); | ||
66 | int r300_mc_wait_for_idle(struct radeon_device *rdev); | ||
67 | int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev); | ||
68 | |||
69 | |||
70 | /* | ||
71 | * rv370,rv380 PCIE GART | ||
72 | */ | ||
73 | void rv370_pcie_gart_tlb_flush(struct radeon_device *rdev) | ||
74 | { | ||
75 | uint32_t tmp; | ||
76 | int i; | ||
77 | |||
78 | /* Workaround HW bug do flush 2 times */ | ||
79 | for (i = 0; i < 2; i++) { | ||
80 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); | ||
81 | WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp | RADEON_PCIE_TX_GART_INVALIDATE_TLB); | ||
82 | (void)RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); | ||
83 | WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); | ||
84 | mb(); | ||
85 | } | ||
86 | } | ||
87 | |||
88 | int rv370_pcie_gart_enable(struct radeon_device *rdev) | ||
89 | { | ||
90 | uint32_t table_addr; | ||
91 | uint32_t tmp; | ||
92 | int r; | ||
93 | |||
94 | /* Initialize common gart structure */ | ||
95 | r = radeon_gart_init(rdev); | ||
96 | if (r) { | ||
97 | return r; | ||
98 | } | ||
99 | r = rv370_debugfs_pcie_gart_info_init(rdev); | ||
100 | if (r) { | ||
101 | DRM_ERROR("Failed to register debugfs file for PCIE gart !\n"); | ||
102 | } | ||
103 | rdev->gart.table_size = rdev->gart.num_gpu_pages * 4; | ||
104 | r = radeon_gart_table_vram_alloc(rdev); | ||
105 | if (r) { | ||
106 | return r; | ||
107 | } | ||
108 | /* discard memory request outside of configured range */ | ||
109 | tmp = RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; | ||
110 | WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); | ||
111 | WREG32_PCIE(RADEON_PCIE_TX_GART_START_LO, rdev->mc.gtt_location); | ||
112 | tmp = rdev->mc.gtt_location + rdev->mc.gtt_size - 4096; | ||
113 | WREG32_PCIE(RADEON_PCIE_TX_GART_END_LO, tmp); | ||
114 | WREG32_PCIE(RADEON_PCIE_TX_GART_START_HI, 0); | ||
115 | WREG32_PCIE(RADEON_PCIE_TX_GART_END_HI, 0); | ||
116 | table_addr = rdev->gart.table_addr; | ||
117 | WREG32_PCIE(RADEON_PCIE_TX_GART_BASE, table_addr); | ||
118 | /* FIXME: setup default page */ | ||
119 | WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_LO, rdev->mc.vram_location); | ||
120 | WREG32_PCIE(RADEON_PCIE_TX_DISCARD_RD_ADDR_HI, 0); | ||
121 | /* Clear error */ | ||
122 | WREG32_PCIE(0x18, 0); | ||
123 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); | ||
124 | tmp |= RADEON_PCIE_TX_GART_EN; | ||
125 | tmp |= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; | ||
126 | WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp); | ||
127 | rv370_pcie_gart_tlb_flush(rdev); | ||
128 | DRM_INFO("PCIE GART of %uM enabled (table at 0x%08X).\n", | ||
129 | rdev->mc.gtt_size >> 20, table_addr); | ||
130 | rdev->gart.ready = true; | ||
131 | return 0; | ||
132 | } | ||
133 | |||
134 | void rv370_pcie_gart_disable(struct radeon_device *rdev) | ||
135 | { | ||
136 | uint32_t tmp; | ||
137 | |||
138 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); | ||
139 | tmp |= RADEON_PCIE_TX_GART_UNMAPPED_ACCESS_DISCARD; | ||
140 | WREG32_PCIE(RADEON_PCIE_TX_GART_CNTL, tmp & ~RADEON_PCIE_TX_GART_EN); | ||
141 | if (rdev->gart.table.vram.robj) { | ||
142 | radeon_object_kunmap(rdev->gart.table.vram.robj); | ||
143 | radeon_object_unpin(rdev->gart.table.vram.robj); | ||
144 | } | ||
145 | } | ||
146 | |||
147 | int rv370_pcie_gart_set_page(struct radeon_device *rdev, int i, uint64_t addr) | ||
148 | { | ||
149 | void __iomem *ptr = (void *)rdev->gart.table.vram.ptr; | ||
150 | |||
151 | if (i < 0 || i > rdev->gart.num_gpu_pages) { | ||
152 | return -EINVAL; | ||
153 | } | ||
154 | addr = (((u32)addr) >> 8) | ((upper_32_bits(addr) & 0xff) << 4) | 0xC; | ||
155 | writel(cpu_to_le32(addr), ((void __iomem *)ptr) + (i * 4)); | ||
156 | return 0; | ||
157 | } | ||
158 | |||
159 | int r300_gart_enable(struct radeon_device *rdev) | ||
160 | { | ||
161 | #if __OS_HAS_AGP | ||
162 | if (rdev->flags & RADEON_IS_AGP) { | ||
163 | if (rdev->family > CHIP_RV350) { | ||
164 | rv370_pcie_gart_disable(rdev); | ||
165 | } else { | ||
166 | r100_pci_gart_disable(rdev); | ||
167 | } | ||
168 | return 0; | ||
169 | } | ||
170 | #endif | ||
171 | if (rdev->flags & RADEON_IS_PCIE) { | ||
172 | rdev->asic->gart_disable = &rv370_pcie_gart_disable; | ||
173 | rdev->asic->gart_tlb_flush = &rv370_pcie_gart_tlb_flush; | ||
174 | rdev->asic->gart_set_page = &rv370_pcie_gart_set_page; | ||
175 | return rv370_pcie_gart_enable(rdev); | ||
176 | } | ||
177 | return r100_pci_gart_enable(rdev); | ||
178 | } | ||
179 | |||
180 | |||
181 | /* | ||
182 | * MC | ||
183 | */ | ||
184 | int r300_mc_init(struct radeon_device *rdev) | ||
185 | { | ||
186 | int r; | ||
187 | |||
188 | if (r100_debugfs_rbbm_init(rdev)) { | ||
189 | DRM_ERROR("Failed to register debugfs file for RBBM !\n"); | ||
190 | } | ||
191 | |||
192 | r300_gpu_init(rdev); | ||
193 | r100_pci_gart_disable(rdev); | ||
194 | if (rdev->flags & RADEON_IS_PCIE) { | ||
195 | rv370_pcie_gart_disable(rdev); | ||
196 | } | ||
197 | |||
198 | /* Setup GPU memory space */ | ||
199 | rdev->mc.vram_location = 0xFFFFFFFFUL; | ||
200 | rdev->mc.gtt_location = 0xFFFFFFFFUL; | ||
201 | if (rdev->flags & RADEON_IS_AGP) { | ||
202 | r = radeon_agp_init(rdev); | ||
203 | if (r) { | ||
204 | printk(KERN_WARNING "[drm] Disabling AGP\n"); | ||
205 | rdev->flags &= ~RADEON_IS_AGP; | ||
206 | rdev->mc.gtt_size = radeon_gart_size * 1024 * 1024; | ||
207 | } else { | ||
208 | rdev->mc.gtt_location = rdev->mc.agp_base; | ||
209 | } | ||
210 | } | ||
211 | r = radeon_mc_setup(rdev); | ||
212 | if (r) { | ||
213 | return r; | ||
214 | } | ||
215 | |||
216 | /* Program GPU memory space */ | ||
217 | r100_mc_disable_clients(rdev); | ||
218 | if (r300_mc_wait_for_idle(rdev)) { | ||
219 | printk(KERN_WARNING "Failed to wait MC idle while " | ||
220 | "programming pipes. Bad things might happen.\n"); | ||
221 | } | ||
222 | r100_mc_setup(rdev); | ||
223 | return 0; | ||
224 | } | ||
225 | |||
226 | void r300_mc_fini(struct radeon_device *rdev) | ||
227 | { | ||
228 | if (rdev->flags & RADEON_IS_PCIE) { | ||
229 | rv370_pcie_gart_disable(rdev); | ||
230 | radeon_gart_table_vram_free(rdev); | ||
231 | } else { | ||
232 | r100_pci_gart_disable(rdev); | ||
233 | radeon_gart_table_ram_free(rdev); | ||
234 | } | ||
235 | radeon_gart_fini(rdev); | ||
236 | } | ||
237 | |||
238 | |||
239 | /* | ||
240 | * Fence emission | ||
241 | */ | ||
242 | void r300_fence_ring_emit(struct radeon_device *rdev, | ||
243 | struct radeon_fence *fence) | ||
244 | { | ||
245 | /* Who ever call radeon_fence_emit should call ring_lock and ask | ||
246 | * for enough space (today caller are ib schedule and buffer move) */ | ||
247 | /* Write SC register so SC & US assert idle */ | ||
248 | radeon_ring_write(rdev, PACKET0(0x43E0, 0)); | ||
249 | radeon_ring_write(rdev, 0); | ||
250 | radeon_ring_write(rdev, PACKET0(0x43E4, 0)); | ||
251 | radeon_ring_write(rdev, 0); | ||
252 | /* Flush 3D cache */ | ||
253 | radeon_ring_write(rdev, PACKET0(0x4E4C, 0)); | ||
254 | radeon_ring_write(rdev, (2 << 0)); | ||
255 | radeon_ring_write(rdev, PACKET0(0x4F18, 0)); | ||
256 | radeon_ring_write(rdev, (1 << 0)); | ||
257 | /* Wait until IDLE & CLEAN */ | ||
258 | radeon_ring_write(rdev, PACKET0(0x1720, 0)); | ||
259 | radeon_ring_write(rdev, (1 << 17) | (1 << 16) | (1 << 9)); | ||
260 | /* Emit fence sequence & fire IRQ */ | ||
261 | radeon_ring_write(rdev, PACKET0(rdev->fence_drv.scratch_reg, 0)); | ||
262 | radeon_ring_write(rdev, fence->seq); | ||
263 | radeon_ring_write(rdev, PACKET0(RADEON_GEN_INT_STATUS, 0)); | ||
264 | radeon_ring_write(rdev, RADEON_SW_INT_FIRE); | ||
265 | } | ||
266 | |||
267 | |||
268 | /* | ||
269 | * Global GPU functions | ||
270 | */ | ||
271 | int r300_copy_dma(struct radeon_device *rdev, | ||
272 | uint64_t src_offset, | ||
273 | uint64_t dst_offset, | ||
274 | unsigned num_pages, | ||
275 | struct radeon_fence *fence) | ||
276 | { | ||
277 | uint32_t size; | ||
278 | uint32_t cur_size; | ||
279 | int i, num_loops; | ||
280 | int r = 0; | ||
281 | |||
282 | /* radeon pitch is /64 */ | ||
283 | size = num_pages << PAGE_SHIFT; | ||
284 | num_loops = DIV_ROUND_UP(size, 0x1FFFFF); | ||
285 | r = radeon_ring_lock(rdev, num_loops * 4 + 64); | ||
286 | if (r) { | ||
287 | DRM_ERROR("radeon: moving bo (%d).\n", r); | ||
288 | return r; | ||
289 | } | ||
290 | /* Must wait for 2D idle & clean before DMA or hangs might happen */ | ||
291 | radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); | ||
292 | radeon_ring_write(rdev, (1 << 16)); | ||
293 | for (i = 0; i < num_loops; i++) { | ||
294 | cur_size = size; | ||
295 | if (cur_size > 0x1FFFFF) { | ||
296 | cur_size = 0x1FFFFF; | ||
297 | } | ||
298 | size -= cur_size; | ||
299 | radeon_ring_write(rdev, PACKET0(0x720, 2)); | ||
300 | radeon_ring_write(rdev, src_offset); | ||
301 | radeon_ring_write(rdev, dst_offset); | ||
302 | radeon_ring_write(rdev, cur_size | (1 << 31) | (1 << 30)); | ||
303 | src_offset += cur_size; | ||
304 | dst_offset += cur_size; | ||
305 | } | ||
306 | radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); | ||
307 | radeon_ring_write(rdev, RADEON_WAIT_DMA_GUI_IDLE); | ||
308 | if (fence) { | ||
309 | r = radeon_fence_emit(rdev, fence); | ||
310 | } | ||
311 | radeon_ring_unlock_commit(rdev); | ||
312 | return r; | ||
313 | } | ||
314 | |||
315 | void r300_ring_start(struct radeon_device *rdev) | ||
316 | { | ||
317 | unsigned gb_tile_config; | ||
318 | int r; | ||
319 | |||
320 | /* Sub pixel 1/12 so we can have 4K rendering according to doc */ | ||
321 | gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16); | ||
322 | switch (rdev->num_gb_pipes) { | ||
323 | case 2: | ||
324 | gb_tile_config |= R300_PIPE_COUNT_R300; | ||
325 | break; | ||
326 | case 3: | ||
327 | gb_tile_config |= R300_PIPE_COUNT_R420_3P; | ||
328 | break; | ||
329 | case 4: | ||
330 | gb_tile_config |= R300_PIPE_COUNT_R420; | ||
331 | break; | ||
332 | case 1: | ||
333 | default: | ||
334 | gb_tile_config |= R300_PIPE_COUNT_RV350; | ||
335 | break; | ||
336 | } | ||
337 | |||
338 | r = radeon_ring_lock(rdev, 64); | ||
339 | if (r) { | ||
340 | return; | ||
341 | } | ||
342 | radeon_ring_write(rdev, PACKET0(RADEON_ISYNC_CNTL, 0)); | ||
343 | radeon_ring_write(rdev, | ||
344 | RADEON_ISYNC_ANY2D_IDLE3D | | ||
345 | RADEON_ISYNC_ANY3D_IDLE2D | | ||
346 | RADEON_ISYNC_WAIT_IDLEGUI | | ||
347 | RADEON_ISYNC_CPSCRATCH_IDLEGUI); | ||
348 | radeon_ring_write(rdev, PACKET0(R300_GB_TILE_CONFIG, 0)); | ||
349 | radeon_ring_write(rdev, gb_tile_config); | ||
350 | radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); | ||
351 | radeon_ring_write(rdev, | ||
352 | RADEON_WAIT_2D_IDLECLEAN | | ||
353 | RADEON_WAIT_3D_IDLECLEAN); | ||
354 | radeon_ring_write(rdev, PACKET0(0x170C, 0)); | ||
355 | radeon_ring_write(rdev, 1 << 31); | ||
356 | radeon_ring_write(rdev, PACKET0(R300_GB_SELECT, 0)); | ||
357 | radeon_ring_write(rdev, 0); | ||
358 | radeon_ring_write(rdev, PACKET0(R300_GB_ENABLE, 0)); | ||
359 | radeon_ring_write(rdev, 0); | ||
360 | radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); | ||
361 | radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); | ||
362 | radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); | ||
363 | radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE); | ||
364 | radeon_ring_write(rdev, PACKET0(RADEON_WAIT_UNTIL, 0)); | ||
365 | radeon_ring_write(rdev, | ||
366 | RADEON_WAIT_2D_IDLECLEAN | | ||
367 | RADEON_WAIT_3D_IDLECLEAN); | ||
368 | radeon_ring_write(rdev, PACKET0(R300_GB_AA_CONFIG, 0)); | ||
369 | radeon_ring_write(rdev, 0); | ||
370 | radeon_ring_write(rdev, PACKET0(R300_RB3D_DSTCACHE_CTLSTAT, 0)); | ||
371 | radeon_ring_write(rdev, R300_RB3D_DC_FLUSH | R300_RB3D_DC_FREE); | ||
372 | radeon_ring_write(rdev, PACKET0(R300_RB3D_ZCACHE_CTLSTAT, 0)); | ||
373 | radeon_ring_write(rdev, R300_ZC_FLUSH | R300_ZC_FREE); | ||
374 | radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS0, 0)); | ||
375 | radeon_ring_write(rdev, | ||
376 | ((6 << R300_MS_X0_SHIFT) | | ||
377 | (6 << R300_MS_Y0_SHIFT) | | ||
378 | (6 << R300_MS_X1_SHIFT) | | ||
379 | (6 << R300_MS_Y1_SHIFT) | | ||
380 | (6 << R300_MS_X2_SHIFT) | | ||
381 | (6 << R300_MS_Y2_SHIFT) | | ||
382 | (6 << R300_MSBD0_Y_SHIFT) | | ||
383 | (6 << R300_MSBD0_X_SHIFT))); | ||
384 | radeon_ring_write(rdev, PACKET0(R300_GB_MSPOS1, 0)); | ||
385 | radeon_ring_write(rdev, | ||
386 | ((6 << R300_MS_X3_SHIFT) | | ||
387 | (6 << R300_MS_Y3_SHIFT) | | ||
388 | (6 << R300_MS_X4_SHIFT) | | ||
389 | (6 << R300_MS_Y4_SHIFT) | | ||
390 | (6 << R300_MS_X5_SHIFT) | | ||
391 | (6 << R300_MS_Y5_SHIFT) | | ||
392 | (6 << R300_MSBD1_SHIFT))); | ||
393 | radeon_ring_write(rdev, PACKET0(R300_GA_ENHANCE, 0)); | ||
394 | radeon_ring_write(rdev, R300_GA_DEADLOCK_CNTL | R300_GA_FASTSYNC_CNTL); | ||
395 | radeon_ring_write(rdev, PACKET0(R300_GA_POLY_MODE, 0)); | ||
396 | radeon_ring_write(rdev, | ||
397 | R300_FRONT_PTYPE_TRIANGE | R300_BACK_PTYPE_TRIANGE); | ||
398 | radeon_ring_write(rdev, PACKET0(R300_GA_ROUND_MODE, 0)); | ||
399 | radeon_ring_write(rdev, | ||
400 | R300_GEOMETRY_ROUND_NEAREST | | ||
401 | R300_COLOR_ROUND_NEAREST); | ||
402 | radeon_ring_unlock_commit(rdev); | ||
403 | } | ||
404 | |||
405 | void r300_errata(struct radeon_device *rdev) | ||
406 | { | ||
407 | rdev->pll_errata = 0; | ||
408 | |||
409 | if (rdev->family == CHIP_R300 && | ||
410 | (RREG32(RADEON_CONFIG_CNTL) & RADEON_CFG_ATI_REV_ID_MASK) == RADEON_CFG_ATI_REV_A11) { | ||
411 | rdev->pll_errata |= CHIP_ERRATA_R300_CG; | ||
412 | } | ||
413 | } | ||
414 | |||
415 | int r300_mc_wait_for_idle(struct radeon_device *rdev) | ||
416 | { | ||
417 | unsigned i; | ||
418 | uint32_t tmp; | ||
419 | |||
420 | for (i = 0; i < rdev->usec_timeout; i++) { | ||
421 | /* read MC_STATUS */ | ||
422 | tmp = RREG32(0x0150); | ||
423 | if (tmp & (1 << 4)) { | ||
424 | return 0; | ||
425 | } | ||
426 | DRM_UDELAY(1); | ||
427 | } | ||
428 | return -1; | ||
429 | } | ||
430 | |||
431 | void r300_gpu_init(struct radeon_device *rdev) | ||
432 | { | ||
433 | uint32_t gb_tile_config, tmp; | ||
434 | |||
435 | r100_hdp_reset(rdev); | ||
436 | /* FIXME: rv380 one pipes ? */ | ||
437 | if ((rdev->family == CHIP_R300) || (rdev->family == CHIP_R350)) { | ||
438 | /* r300,r350 */ | ||
439 | rdev->num_gb_pipes = 2; | ||
440 | } else { | ||
441 | /* rv350,rv370,rv380 */ | ||
442 | rdev->num_gb_pipes = 1; | ||
443 | } | ||
444 | gb_tile_config = (R300_ENABLE_TILING | R300_TILE_SIZE_16); | ||
445 | switch (rdev->num_gb_pipes) { | ||
446 | case 2: | ||
447 | gb_tile_config |= R300_PIPE_COUNT_R300; | ||
448 | break; | ||
449 | case 3: | ||
450 | gb_tile_config |= R300_PIPE_COUNT_R420_3P; | ||
451 | break; | ||
452 | case 4: | ||
453 | gb_tile_config |= R300_PIPE_COUNT_R420; | ||
454 | break; | ||
455 | case 1: | ||
456 | default: | ||
457 | gb_tile_config |= R300_PIPE_COUNT_RV350; | ||
458 | break; | ||
459 | } | ||
460 | WREG32(R300_GB_TILE_CONFIG, gb_tile_config); | ||
461 | |||
462 | if (r100_gui_wait_for_idle(rdev)) { | ||
463 | printk(KERN_WARNING "Failed to wait GUI idle while " | ||
464 | "programming pipes. Bad things might happen.\n"); | ||
465 | } | ||
466 | |||
467 | tmp = RREG32(0x170C); | ||
468 | WREG32(0x170C, tmp | (1 << 31)); | ||
469 | |||
470 | WREG32(R300_RB2D_DSTCACHE_MODE, | ||
471 | R300_DC_AUTOFLUSH_ENABLE | | ||
472 | R300_DC_DC_DISABLE_IGNORE_PE); | ||
473 | |||
474 | if (r100_gui_wait_for_idle(rdev)) { | ||
475 | printk(KERN_WARNING "Failed to wait GUI idle while " | ||
476 | "programming pipes. Bad things might happen.\n"); | ||
477 | } | ||
478 | if (r300_mc_wait_for_idle(rdev)) { | ||
479 | printk(KERN_WARNING "Failed to wait MC idle while " | ||
480 | "programming pipes. Bad things might happen.\n"); | ||
481 | } | ||
482 | DRM_INFO("radeon: %d pipes initialized.\n", rdev->num_gb_pipes); | ||
483 | } | ||
484 | |||
485 | int r300_ga_reset(struct radeon_device *rdev) | ||
486 | { | ||
487 | uint32_t tmp; | ||
488 | bool reinit_cp; | ||
489 | int i; | ||
490 | |||
491 | reinit_cp = rdev->cp.ready; | ||
492 | rdev->cp.ready = false; | ||
493 | for (i = 0; i < rdev->usec_timeout; i++) { | ||
494 | WREG32(RADEON_CP_CSQ_MODE, 0); | ||
495 | WREG32(RADEON_CP_CSQ_CNTL, 0); | ||
496 | WREG32(RADEON_RBBM_SOFT_RESET, 0x32005); | ||
497 | (void)RREG32(RADEON_RBBM_SOFT_RESET); | ||
498 | udelay(200); | ||
499 | WREG32(RADEON_RBBM_SOFT_RESET, 0); | ||
500 | /* Wait to prevent race in RBBM_STATUS */ | ||
501 | mdelay(1); | ||
502 | tmp = RREG32(RADEON_RBBM_STATUS); | ||
503 | if (tmp & ((1 << 20) | (1 << 26))) { | ||
504 | DRM_ERROR("VAP & CP still busy (RBBM_STATUS=0x%08X)", tmp); | ||
505 | /* GA still busy soft reset it */ | ||
506 | WREG32(0x429C, 0x200); | ||
507 | WREG32(R300_VAP_PVS_STATE_FLUSH_REG, 0); | ||
508 | WREG32(0x43E0, 0); | ||
509 | WREG32(0x43E4, 0); | ||
510 | WREG32(0x24AC, 0); | ||
511 | } | ||
512 | /* Wait to prevent race in RBBM_STATUS */ | ||
513 | mdelay(1); | ||
514 | tmp = RREG32(RADEON_RBBM_STATUS); | ||
515 | if (!(tmp & ((1 << 20) | (1 << 26)))) { | ||
516 | break; | ||
517 | } | ||
518 | } | ||
519 | for (i = 0; i < rdev->usec_timeout; i++) { | ||
520 | tmp = RREG32(RADEON_RBBM_STATUS); | ||
521 | if (!(tmp & ((1 << 20) | (1 << 26)))) { | ||
522 | DRM_INFO("GA reset succeed (RBBM_STATUS=0x%08X)\n", | ||
523 | tmp); | ||
524 | if (reinit_cp) { | ||
525 | return r100_cp_init(rdev, rdev->cp.ring_size); | ||
526 | } | ||
527 | return 0; | ||
528 | } | ||
529 | DRM_UDELAY(1); | ||
530 | } | ||
531 | tmp = RREG32(RADEON_RBBM_STATUS); | ||
532 | DRM_ERROR("Failed to reset GA ! (RBBM_STATUS=0x%08X)\n", tmp); | ||
533 | return -1; | ||
534 | } | ||
535 | |||
536 | int r300_gpu_reset(struct radeon_device *rdev) | ||
537 | { | ||
538 | uint32_t status; | ||
539 | |||
540 | /* reset order likely matter */ | ||
541 | status = RREG32(RADEON_RBBM_STATUS); | ||
542 | /* reset HDP */ | ||
543 | r100_hdp_reset(rdev); | ||
544 | /* reset rb2d */ | ||
545 | if (status & ((1 << 17) | (1 << 18) | (1 << 27))) { | ||
546 | r100_rb2d_reset(rdev); | ||
547 | } | ||
548 | /* reset GA */ | ||
549 | if (status & ((1 << 20) | (1 << 26))) { | ||
550 | r300_ga_reset(rdev); | ||
551 | } | ||
552 | /* reset CP */ | ||
553 | status = RREG32(RADEON_RBBM_STATUS); | ||
554 | if (status & (1 << 16)) { | ||
555 | r100_cp_reset(rdev); | ||
556 | } | ||
557 | /* Check if GPU is idle */ | ||
558 | status = RREG32(RADEON_RBBM_STATUS); | ||
559 | if (status & (1 << 31)) { | ||
560 | DRM_ERROR("Failed to reset GPU (RBBM_STATUS=0x%08X)\n", status); | ||
561 | return -1; | ||
562 | } | ||
563 | DRM_INFO("GPU reset succeed (RBBM_STATUS=0x%08X)\n", status); | ||
564 | return 0; | ||
565 | } | ||
566 | |||
567 | |||
568 | /* | ||
569 | * r300,r350,rv350,rv380 VRAM info | ||
570 | */ | ||
571 | void r300_vram_info(struct radeon_device *rdev) | ||
572 | { | ||
573 | uint32_t tmp; | ||
574 | |||
575 | /* DDR for all card after R300 & IGP */ | ||
576 | rdev->mc.vram_is_ddr = true; | ||
577 | tmp = RREG32(RADEON_MEM_CNTL); | ||
578 | if (tmp & R300_MEM_NUM_CHANNELS_MASK) { | ||
579 | rdev->mc.vram_width = 128; | ||
580 | } else { | ||
581 | rdev->mc.vram_width = 64; | ||
582 | } | ||
583 | rdev->mc.vram_size = RREG32(RADEON_CONFIG_MEMSIZE); | ||
584 | |||
585 | rdev->mc.aper_base = drm_get_resource_start(rdev->ddev, 0); | ||
586 | rdev->mc.aper_size = drm_get_resource_len(rdev->ddev, 0); | ||
587 | } | ||
588 | |||
589 | |||
590 | /* | ||
591 | * Indirect registers accessor | ||
592 | */ | ||
593 | uint32_t rv370_pcie_rreg(struct radeon_device *rdev, uint32_t reg) | ||
594 | { | ||
595 | uint32_t r; | ||
596 | |||
597 | WREG8(RADEON_PCIE_INDEX, ((reg) & 0xff)); | ||
598 | (void)RREG32(RADEON_PCIE_INDEX); | ||
599 | r = RREG32(RADEON_PCIE_DATA); | ||
600 | return r; | ||
601 | } | ||
602 | |||
603 | void rv370_pcie_wreg(struct radeon_device *rdev, uint32_t reg, uint32_t v) | ||
604 | { | ||
605 | WREG8(RADEON_PCIE_INDEX, ((reg) & 0xff)); | ||
606 | (void)RREG32(RADEON_PCIE_INDEX); | ||
607 | WREG32(RADEON_PCIE_DATA, (v)); | ||
608 | (void)RREG32(RADEON_PCIE_DATA); | ||
609 | } | ||
610 | |||
611 | /* | ||
612 | * PCIE Lanes | ||
613 | */ | ||
614 | |||
615 | void rv370_set_pcie_lanes(struct radeon_device *rdev, int lanes) | ||
616 | { | ||
617 | uint32_t link_width_cntl, mask; | ||
618 | |||
619 | if (rdev->flags & RADEON_IS_IGP) | ||
620 | return; | ||
621 | |||
622 | if (!(rdev->flags & RADEON_IS_PCIE)) | ||
623 | return; | ||
624 | |||
625 | /* FIXME wait for idle */ | ||
626 | |||
627 | switch (lanes) { | ||
628 | case 0: | ||
629 | mask = RADEON_PCIE_LC_LINK_WIDTH_X0; | ||
630 | break; | ||
631 | case 1: | ||
632 | mask = RADEON_PCIE_LC_LINK_WIDTH_X1; | ||
633 | break; | ||
634 | case 2: | ||
635 | mask = RADEON_PCIE_LC_LINK_WIDTH_X2; | ||
636 | break; | ||
637 | case 4: | ||
638 | mask = RADEON_PCIE_LC_LINK_WIDTH_X4; | ||
639 | break; | ||
640 | case 8: | ||
641 | mask = RADEON_PCIE_LC_LINK_WIDTH_X8; | ||
642 | break; | ||
643 | case 12: | ||
644 | mask = RADEON_PCIE_LC_LINK_WIDTH_X12; | ||
645 | break; | ||
646 | case 16: | ||
647 | default: | ||
648 | mask = RADEON_PCIE_LC_LINK_WIDTH_X16; | ||
649 | break; | ||
650 | } | ||
651 | |||
652 | link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); | ||
653 | |||
654 | if ((link_width_cntl & RADEON_PCIE_LC_LINK_WIDTH_RD_MASK) == | ||
655 | (mask << RADEON_PCIE_LC_LINK_WIDTH_RD_SHIFT)) | ||
656 | return; | ||
657 | |||
658 | link_width_cntl &= ~(RADEON_PCIE_LC_LINK_WIDTH_MASK | | ||
659 | RADEON_PCIE_LC_RECONFIG_NOW | | ||
660 | RADEON_PCIE_LC_RECONFIG_LATER | | ||
661 | RADEON_PCIE_LC_SHORT_RECONFIG_EN); | ||
662 | link_width_cntl |= mask; | ||
663 | WREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl); | ||
664 | WREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL, (link_width_cntl | | ||
665 | RADEON_PCIE_LC_RECONFIG_NOW)); | ||
666 | |||
667 | /* wait for lane set to complete */ | ||
668 | link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); | ||
669 | while (link_width_cntl == 0xffffffff) | ||
670 | link_width_cntl = RREG32_PCIE(RADEON_PCIE_LC_LINK_WIDTH_CNTL); | ||
671 | |||
672 | } | ||
673 | |||
674 | |||
675 | /* | ||
676 | * Debugfs info | ||
677 | */ | ||
678 | #if defined(CONFIG_DEBUG_FS) | ||
679 | static int rv370_debugfs_pcie_gart_info(struct seq_file *m, void *data) | ||
680 | { | ||
681 | struct drm_info_node *node = (struct drm_info_node *) m->private; | ||
682 | struct drm_device *dev = node->minor->dev; | ||
683 | struct radeon_device *rdev = dev->dev_private; | ||
684 | uint32_t tmp; | ||
685 | |||
686 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_CNTL); | ||
687 | seq_printf(m, "PCIE_TX_GART_CNTL 0x%08x\n", tmp); | ||
688 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_BASE); | ||
689 | seq_printf(m, "PCIE_TX_GART_BASE 0x%08x\n", tmp); | ||
690 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_START_LO); | ||
691 | seq_printf(m, "PCIE_TX_GART_START_LO 0x%08x\n", tmp); | ||
692 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_START_HI); | ||
693 | seq_printf(m, "PCIE_TX_GART_START_HI 0x%08x\n", tmp); | ||
694 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_END_LO); | ||
695 | seq_printf(m, "PCIE_TX_GART_END_LO 0x%08x\n", tmp); | ||
696 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_END_HI); | ||
697 | seq_printf(m, "PCIE_TX_GART_END_HI 0x%08x\n", tmp); | ||
698 | tmp = RREG32_PCIE(RADEON_PCIE_TX_GART_ERROR); | ||
699 | seq_printf(m, "PCIE_TX_GART_ERROR 0x%08x\n", tmp); | ||
700 | return 0; | ||
701 | } | ||
702 | |||
703 | static struct drm_info_list rv370_pcie_gart_info_list[] = { | ||
704 | {"rv370_pcie_gart_info", rv370_debugfs_pcie_gart_info, 0, NULL}, | ||
705 | }; | ||
706 | #endif | ||
707 | |||
708 | int rv370_debugfs_pcie_gart_info_init(struct radeon_device *rdev) | ||
709 | { | ||
710 | #if defined(CONFIG_DEBUG_FS) | ||
711 | return radeon_debugfs_add_files(rdev, rv370_pcie_gart_info_list, 1); | ||
712 | #else | ||
713 | return 0; | ||
714 | #endif | ||
715 | } | ||
716 | |||
717 | |||
718 | /* | ||
719 | * CS functions | ||
720 | */ | ||
721 | struct r300_cs_track_cb { | ||
722 | struct radeon_object *robj; | ||
723 | unsigned pitch; | ||
724 | unsigned cpp; | ||
725 | unsigned offset; | ||
726 | }; | ||
727 | |||
728 | struct r300_cs_track { | ||
729 | unsigned num_cb; | ||
730 | unsigned maxy; | ||
731 | struct r300_cs_track_cb cb[4]; | ||
732 | struct r300_cs_track_cb zb; | ||
733 | bool z_enabled; | ||
734 | }; | ||
735 | |||
736 | int r300_cs_track_check(struct radeon_device *rdev, struct r300_cs_track *track) | ||
737 | { | ||
738 | unsigned i; | ||
739 | unsigned long size; | ||
740 | |||
741 | for (i = 0; i < track->num_cb; i++) { | ||
742 | if (track->cb[i].robj == NULL) { | ||
743 | DRM_ERROR("[drm] No buffer for color buffer %d !\n", i); | ||
744 | return -EINVAL; | ||
745 | } | ||
746 | size = track->cb[i].pitch * track->cb[i].cpp * track->maxy; | ||
747 | size += track->cb[i].offset; | ||
748 | if (size > radeon_object_size(track->cb[i].robj)) { | ||
749 | DRM_ERROR("[drm] Buffer too small for color buffer %d " | ||
750 | "(need %lu have %lu) !\n", i, size, | ||
751 | radeon_object_size(track->cb[i].robj)); | ||
752 | DRM_ERROR("[drm] color buffer %d (%u %u %u %u)\n", | ||
753 | i, track->cb[i].pitch, track->cb[i].cpp, | ||
754 | track->cb[i].offset, track->maxy); | ||
755 | return -EINVAL; | ||
756 | } | ||
757 | } | ||
758 | if (track->z_enabled) { | ||
759 | if (track->zb.robj == NULL) { | ||
760 | DRM_ERROR("[drm] No buffer for z buffer !\n"); | ||
761 | return -EINVAL; | ||
762 | } | ||
763 | size = track->zb.pitch * track->zb.cpp * track->maxy; | ||
764 | size += track->zb.offset; | ||
765 | if (size > radeon_object_size(track->zb.robj)) { | ||
766 | DRM_ERROR("[drm] Buffer too small for z buffer " | ||
767 | "(need %lu have %lu) !\n", size, | ||
768 | radeon_object_size(track->zb.robj)); | ||
769 | return -EINVAL; | ||
770 | } | ||
771 | } | ||
772 | return 0; | ||
773 | } | ||
774 | |||
775 | static inline void r300_cs_track_clear(struct r300_cs_track *track) | ||
776 | { | ||
777 | unsigned i; | ||
778 | |||
779 | track->num_cb = 4; | ||
780 | track->maxy = 4096; | ||
781 | for (i = 0; i < track->num_cb; i++) { | ||
782 | track->cb[i].robj = NULL; | ||
783 | track->cb[i].pitch = 8192; | ||
784 | track->cb[i].cpp = 16; | ||
785 | track->cb[i].offset = 0; | ||
786 | } | ||
787 | track->z_enabled = true; | ||
788 | track->zb.robj = NULL; | ||
789 | track->zb.pitch = 8192; | ||
790 | track->zb.cpp = 4; | ||
791 | track->zb.offset = 0; | ||
792 | } | ||
793 | |||
794 | static unsigned r300_auth_reg[] = { | ||
795 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
796 | 0xFFFFFFBF, 0xFFFFFFFF, 0xFFFFFFBF, 0xFFFFFFFF, | ||
797 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
798 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
799 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
800 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
801 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
802 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
803 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
804 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
805 | 0x17FF1FFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFF30FFBF, | ||
806 | 0xFFFFFFF8, 0xC3E6FFFF, 0xFFFFF6DF, 0xFFFFFFFF, | ||
807 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
808 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
809 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFF03F, | ||
810 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
811 | 0xFFFFFFFF, 0xFFFFCFCC, 0xF00E9FFF, 0x007C0000, | ||
812 | 0xF0000078, 0xFF000009, 0xFFFFFFFF, 0xFFFFFFFF, | ||
813 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
814 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
815 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
816 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
817 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
818 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
819 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
820 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
821 | 0xFFFFF7FF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
822 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
823 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
824 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
825 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
826 | 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, | ||
827 | 0xFFFFFC78, 0xFFFFFFFF, 0xFFFFFFFC, 0xFFFFFFFF, | ||
828 | 0x38FF8F50, 0xFFF88082, 0xF000000C, 0xFAE009FF, | ||
829 | 0x00000000, 0x00000000, 0xFFFF0000, 0x00000000, | ||
830 | 0x00000000, 0x0000C100, 0x00000000, 0x00000000, | ||
831 | 0x00000000, 0x00000000, 0x00000000, 0x00000000, | ||
832 | 0x00000000, 0xFFFF0000, 0xFFFFFFFF, 0xFF80FFFF, | ||
833 | 0x00000000, 0x00000000, 0x00000000, 0x00000000, | ||
834 | 0x0003FC01, 0xFFFFFFF8, 0xFE800B19, | ||
835 | }; | ||
836 | |||
837 | static int r300_packet0_check(struct radeon_cs_parser *p, | ||
838 | struct radeon_cs_packet *pkt, | ||
839 | unsigned idx, unsigned reg) | ||
840 | { | ||
841 | struct radeon_cs_chunk *ib_chunk; | ||
842 | struct radeon_cs_reloc *reloc; | ||
843 | struct r300_cs_track *track; | ||
844 | volatile uint32_t *ib; | ||
845 | uint32_t tmp; | ||
846 | unsigned i; | ||
847 | int r; | ||
848 | |||
849 | ib = p->ib->ptr; | ||
850 | ib_chunk = &p->chunks[p->chunk_ib_idx]; | ||
851 | track = (struct r300_cs_track *)p->track; | ||
852 | switch (reg) { | ||
853 | case RADEON_DST_PITCH_OFFSET: | ||
854 | case RADEON_SRC_PITCH_OFFSET: | ||
855 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
856 | if (r) { | ||
857 | DRM_ERROR("No reloc for ib[%d]=0x%04X\n", | ||
858 | idx, reg); | ||
859 | r100_cs_dump_packet(p, pkt); | ||
860 | return r; | ||
861 | } | ||
862 | tmp = ib_chunk->kdata[idx] & 0x003fffff; | ||
863 | tmp += (((u32)reloc->lobj.gpu_offset) >> 10); | ||
864 | ib[idx] = (ib_chunk->kdata[idx] & 0xffc00000) | tmp; | ||
865 | break; | ||
866 | case R300_RB3D_COLOROFFSET0: | ||
867 | case R300_RB3D_COLOROFFSET1: | ||
868 | case R300_RB3D_COLOROFFSET2: | ||
869 | case R300_RB3D_COLOROFFSET3: | ||
870 | i = (reg - R300_RB3D_COLOROFFSET0) >> 2; | ||
871 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
872 | if (r) { | ||
873 | DRM_ERROR("No reloc for ib[%d]=0x%04X\n", | ||
874 | idx, reg); | ||
875 | r100_cs_dump_packet(p, pkt); | ||
876 | return r; | ||
877 | } | ||
878 | track->cb[i].robj = reloc->robj; | ||
879 | track->cb[i].offset = ib_chunk->kdata[idx]; | ||
880 | ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); | ||
881 | break; | ||
882 | case R300_ZB_DEPTHOFFSET: | ||
883 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
884 | if (r) { | ||
885 | DRM_ERROR("No reloc for ib[%d]=0x%04X\n", | ||
886 | idx, reg); | ||
887 | r100_cs_dump_packet(p, pkt); | ||
888 | return r; | ||
889 | } | ||
890 | track->zb.robj = reloc->robj; | ||
891 | track->zb.offset = ib_chunk->kdata[idx]; | ||
892 | ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); | ||
893 | break; | ||
894 | case R300_TX_OFFSET_0: | ||
895 | case R300_TX_OFFSET_0+4: | ||
896 | case R300_TX_OFFSET_0+8: | ||
897 | case R300_TX_OFFSET_0+12: | ||
898 | case R300_TX_OFFSET_0+16: | ||
899 | case R300_TX_OFFSET_0+20: | ||
900 | case R300_TX_OFFSET_0+24: | ||
901 | case R300_TX_OFFSET_0+28: | ||
902 | case R300_TX_OFFSET_0+32: | ||
903 | case R300_TX_OFFSET_0+36: | ||
904 | case R300_TX_OFFSET_0+40: | ||
905 | case R300_TX_OFFSET_0+44: | ||
906 | case R300_TX_OFFSET_0+48: | ||
907 | case R300_TX_OFFSET_0+52: | ||
908 | case R300_TX_OFFSET_0+56: | ||
909 | case R300_TX_OFFSET_0+60: | ||
910 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
911 | if (r) { | ||
912 | DRM_ERROR("No reloc for ib[%d]=0x%04X\n", | ||
913 | idx, reg); | ||
914 | r100_cs_dump_packet(p, pkt); | ||
915 | return r; | ||
916 | } | ||
917 | ib[idx] = ib_chunk->kdata[idx] + ((u32)reloc->lobj.gpu_offset); | ||
918 | break; | ||
919 | /* Tracked registers */ | ||
920 | case 0x43E4: | ||
921 | /* SC_SCISSOR1 */ | ||
922 | |||
923 | track->maxy = ((ib_chunk->kdata[idx] >> 13) & 0x1FFF) + 1; | ||
924 | if (p->rdev->family < CHIP_RV515) { | ||
925 | track->maxy -= 1440; | ||
926 | } | ||
927 | break; | ||
928 | case 0x4E00: | ||
929 | /* RB3D_CCTL */ | ||
930 | track->num_cb = ((ib_chunk->kdata[idx] >> 5) & 0x3) + 1; | ||
931 | break; | ||
932 | case 0x4E38: | ||
933 | case 0x4E3C: | ||
934 | case 0x4E40: | ||
935 | case 0x4E44: | ||
936 | /* RB3D_COLORPITCH0 */ | ||
937 | /* RB3D_COLORPITCH1 */ | ||
938 | /* RB3D_COLORPITCH2 */ | ||
939 | /* RB3D_COLORPITCH3 */ | ||
940 | i = (reg - 0x4E38) >> 2; | ||
941 | track->cb[i].pitch = ib_chunk->kdata[idx] & 0x3FFE; | ||
942 | switch (((ib_chunk->kdata[idx] >> 21) & 0xF)) { | ||
943 | case 9: | ||
944 | case 11: | ||
945 | case 12: | ||
946 | track->cb[i].cpp = 1; | ||
947 | break; | ||
948 | case 3: | ||
949 | case 4: | ||
950 | case 13: | ||
951 | case 15: | ||
952 | track->cb[i].cpp = 2; | ||
953 | break; | ||
954 | case 6: | ||
955 | track->cb[i].cpp = 4; | ||
956 | break; | ||
957 | case 10: | ||
958 | track->cb[i].cpp = 8; | ||
959 | break; | ||
960 | case 7: | ||
961 | track->cb[i].cpp = 16; | ||
962 | break; | ||
963 | default: | ||
964 | DRM_ERROR("Invalid color buffer format (%d) !\n", | ||
965 | ((ib_chunk->kdata[idx] >> 21) & 0xF)); | ||
966 | return -EINVAL; | ||
967 | } | ||
968 | break; | ||
969 | case 0x4F00: | ||
970 | /* ZB_CNTL */ | ||
971 | if (ib_chunk->kdata[idx] & 2) { | ||
972 | track->z_enabled = true; | ||
973 | } else { | ||
974 | track->z_enabled = false; | ||
975 | } | ||
976 | break; | ||
977 | case 0x4F10: | ||
978 | /* ZB_FORMAT */ | ||
979 | switch ((ib_chunk->kdata[idx] & 0xF)) { | ||
980 | case 0: | ||
981 | case 1: | ||
982 | track->zb.cpp = 2; | ||
983 | break; | ||
984 | case 2: | ||
985 | track->zb.cpp = 4; | ||
986 | break; | ||
987 | default: | ||
988 | DRM_ERROR("Invalid z buffer format (%d) !\n", | ||
989 | (ib_chunk->kdata[idx] & 0xF)); | ||
990 | return -EINVAL; | ||
991 | } | ||
992 | break; | ||
993 | case 0x4F24: | ||
994 | /* ZB_DEPTHPITCH */ | ||
995 | track->zb.pitch = ib_chunk->kdata[idx] & 0x3FFC; | ||
996 | break; | ||
997 | default: | ||
998 | printk(KERN_ERR "Forbidden register 0x%04X in cs at %d\n", reg, idx); | ||
999 | return -EINVAL; | ||
1000 | } | ||
1001 | return 0; | ||
1002 | } | ||
1003 | |||
1004 | static int r300_packet3_check(struct radeon_cs_parser *p, | ||
1005 | struct radeon_cs_packet *pkt) | ||
1006 | { | ||
1007 | struct radeon_cs_chunk *ib_chunk; | ||
1008 | struct radeon_cs_reloc *reloc; | ||
1009 | struct r300_cs_track *track; | ||
1010 | volatile uint32_t *ib; | ||
1011 | unsigned idx; | ||
1012 | unsigned i, c; | ||
1013 | int r; | ||
1014 | |||
1015 | ib = p->ib->ptr; | ||
1016 | ib_chunk = &p->chunks[p->chunk_ib_idx]; | ||
1017 | idx = pkt->idx + 1; | ||
1018 | track = (struct r300_cs_track *)p->track; | ||
1019 | switch (pkt->opcode) { | ||
1020 | case PACKET3_3D_LOAD_VBPNTR: | ||
1021 | c = ib_chunk->kdata[idx++]; | ||
1022 | for (i = 0; i < (c - 1); i += 2, idx += 3) { | ||
1023 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
1024 | if (r) { | ||
1025 | DRM_ERROR("No reloc for packet3 %d\n", | ||
1026 | pkt->opcode); | ||
1027 | r100_cs_dump_packet(p, pkt); | ||
1028 | return r; | ||
1029 | } | ||
1030 | ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); | ||
1031 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
1032 | if (r) { | ||
1033 | DRM_ERROR("No reloc for packet3 %d\n", | ||
1034 | pkt->opcode); | ||
1035 | r100_cs_dump_packet(p, pkt); | ||
1036 | return r; | ||
1037 | } | ||
1038 | ib[idx+2] = ib_chunk->kdata[idx+2] + ((u32)reloc->lobj.gpu_offset); | ||
1039 | } | ||
1040 | if (c & 1) { | ||
1041 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
1042 | if (r) { | ||
1043 | DRM_ERROR("No reloc for packet3 %d\n", | ||
1044 | pkt->opcode); | ||
1045 | r100_cs_dump_packet(p, pkt); | ||
1046 | return r; | ||
1047 | } | ||
1048 | ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); | ||
1049 | } | ||
1050 | break; | ||
1051 | case PACKET3_INDX_BUFFER: | ||
1052 | r = r100_cs_packet_next_reloc(p, &reloc); | ||
1053 | if (r) { | ||
1054 | DRM_ERROR("No reloc for packet3 %d\n", pkt->opcode); | ||
1055 | r100_cs_dump_packet(p, pkt); | ||
1056 | return r; | ||
1057 | } | ||
1058 | ib[idx+1] = ib_chunk->kdata[idx+1] + ((u32)reloc->lobj.gpu_offset); | ||
1059 | break; | ||
1060 | /* Draw packet */ | ||
1061 | case PACKET3_3D_DRAW_VBUF: | ||
1062 | case PACKET3_3D_DRAW_IMMD: | ||
1063 | case PACKET3_3D_DRAW_INDX: | ||
1064 | case PACKET3_3D_DRAW_VBUF_2: | ||
1065 | case PACKET3_3D_DRAW_IMMD_2: | ||
1066 | case PACKET3_3D_DRAW_INDX_2: | ||
1067 | r = r300_cs_track_check(p->rdev, track); | ||
1068 | if (r) { | ||
1069 | return r; | ||
1070 | } | ||
1071 | break; | ||
1072 | case PACKET3_NOP: | ||
1073 | break; | ||
1074 | default: | ||
1075 | DRM_ERROR("Packet3 opcode %x not supported\n", pkt->opcode); | ||
1076 | return -EINVAL; | ||
1077 | } | ||
1078 | return 0; | ||
1079 | } | ||
1080 | |||
1081 | int r300_cs_parse(struct radeon_cs_parser *p) | ||
1082 | { | ||
1083 | struct radeon_cs_packet pkt; | ||
1084 | struct r300_cs_track track; | ||
1085 | int r; | ||
1086 | |||
1087 | r300_cs_track_clear(&track); | ||
1088 | p->track = &track; | ||
1089 | do { | ||
1090 | r = r100_cs_packet_parse(p, &pkt, p->idx); | ||
1091 | if (r) { | ||
1092 | return r; | ||
1093 | } | ||
1094 | p->idx += pkt.count + 2; | ||
1095 | switch (pkt.type) { | ||
1096 | case PACKET_TYPE0: | ||
1097 | r = r100_cs_parse_packet0(p, &pkt, | ||
1098 | r300_auth_reg, | ||
1099 | ARRAY_SIZE(r300_auth_reg), | ||
1100 | &r300_packet0_check); | ||
1101 | break; | ||
1102 | case PACKET_TYPE2: | ||
1103 | break; | ||
1104 | case PACKET_TYPE3: | ||
1105 | r = r300_packet3_check(p, &pkt); | ||
1106 | break; | ||
1107 | default: | ||
1108 | DRM_ERROR("Unknown packet type %d !\n", pkt.type); | ||
1109 | return -EINVAL; | ||
1110 | } | ||
1111 | if (r) { | ||
1112 | return r; | ||
1113 | } | ||
1114 | } while (p->idx < p->chunks[p->chunk_ib_idx].length_dw); | ||
1115 | return 0; | ||
1116 | } | ||