1/*
2 * Copyright 2013 Advanced Micro Devices, Inc.
3 * All Rights Reserved.
4 *
5 * Permission is hereby granted, free of charge, to any person obtaining a
6 * copy of this software and associated documentation files (the
7 * "Software"), to deal in the Software without restriction, including
8 * without limitation the rights to use, copy, modify, merge, publish,
9 * distribute, sub license, and/or sell copies of the Software, and to
10 * permit persons to whom the Software is furnished to do so, subject to
11 * the following conditions:
12 *
13 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
14 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
15 * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
16 * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
17 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
18 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
19 * USE OR OTHER DEALINGS IN THE SOFTWARE.
20 *
21 * The above copyright notice and this permission notice (including the
22 * next paragraph) shall be included in all copies or substantial portions
23 * of the Software.
24 *
25 * Authors: Christian König <christian.koenig@amd.com>
26 */
27
28#include <linux/firmware.h>
29
30#include "radeon.h"
31#include "radeon_asic.h"
32#include "sid.h"
33
34#define VCE_V1_0_FW_SIZE	(256 * 1024)
35#define VCE_V1_0_STACK_SIZE	(64 * 1024)
36#define VCE_V1_0_DATA_SIZE	(7808 * (RADEON_MAX_VCE_HANDLES + 1))
37
38struct vce_v1_0_fw_signature
39{
40	int32_t off;
41	uint32_t len;
42	int32_t num;
43	struct {
44		uint32_t chip_id;
45		uint32_t keyselect;
46		uint32_t nonce[4];
47		uint32_t sigval[4];
48	} val[8];
49};
50
51/**
52 * vce_v1_0_get_rptr - get read pointer
53 *
54 * @rdev: radeon_device pointer
55 * @ring: radeon_ring pointer
56 *
57 * Returns the current hardware read pointer
58 */
59uint32_t vce_v1_0_get_rptr(struct radeon_device *rdev,
60			   struct radeon_ring *ring)
61{
62	if (ring->idx == TN_RING_TYPE_VCE1_INDEX)
63		return RREG32(VCE_RB_RPTR);
64	else
65		return RREG32(VCE_RB_RPTR2);
66}
67
68/**
69 * vce_v1_0_get_wptr - get write pointer
70 *
71 * @rdev: radeon_device pointer
72 * @ring: radeon_ring pointer
73 *
74 * Returns the current hardware write pointer
75 */
76uint32_t vce_v1_0_get_wptr(struct radeon_device *rdev,
77			   struct radeon_ring *ring)
78{
79	if (ring->idx == TN_RING_TYPE_VCE1_INDEX)
80		return RREG32(VCE_RB_WPTR);
81	else
82		return RREG32(VCE_RB_WPTR2);
83}
84
85/**
86 * vce_v1_0_set_wptr - set write pointer
87 *
88 * @rdev: radeon_device pointer
89 * @ring: radeon_ring pointer
90 *
91 * Commits the write pointer to the hardware
92 */
93void vce_v1_0_set_wptr(struct radeon_device *rdev,
94		       struct radeon_ring *ring)
95{
96	if (ring->idx == TN_RING_TYPE_VCE1_INDEX)
97		WREG32(VCE_RB_WPTR, ring->wptr);
98	else
99		WREG32(VCE_RB_WPTR2, ring->wptr);
100}
101
102void vce_v1_0_enable_mgcg(struct radeon_device *rdev, bool enable)
103{
104	u32 tmp;
105
106	if (enable && (rdev->cg_flags & RADEON_CG_SUPPORT_VCE_MGCG)) {
107		tmp = RREG32(VCE_CLOCK_GATING_A);
108		tmp |= CGC_DYN_CLOCK_MODE;
109		WREG32(VCE_CLOCK_GATING_A, tmp);
110
111		tmp = RREG32(VCE_UENC_CLOCK_GATING);
112		tmp &= ~0x1ff000;
113		tmp |= 0xff800000;
114		WREG32(VCE_UENC_CLOCK_GATING, tmp);
115
116		tmp = RREG32(VCE_UENC_REG_CLOCK_GATING);
117		tmp &= ~0x3ff;
118		WREG32(VCE_UENC_REG_CLOCK_GATING, tmp);
119	} else {
120		tmp = RREG32(VCE_CLOCK_GATING_A);
121		tmp &= ~CGC_DYN_CLOCK_MODE;
122		WREG32(VCE_CLOCK_GATING_A, tmp);
123
124		tmp = RREG32(VCE_UENC_CLOCK_GATING);
125		tmp |= 0x1ff000;
126		tmp &= ~0xff800000;
127		WREG32(VCE_UENC_CLOCK_GATING, tmp);
128
129		tmp = RREG32(VCE_UENC_REG_CLOCK_GATING);
130		tmp |= 0x3ff;
131		WREG32(VCE_UENC_REG_CLOCK_GATING, tmp);
132	}
133}
134
135static void vce_v1_0_init_cg(struct radeon_device *rdev)
136{
137	u32 tmp;
138
139	tmp = RREG32(VCE_CLOCK_GATING_A);
140	tmp |= CGC_DYN_CLOCK_MODE;
141	WREG32(VCE_CLOCK_GATING_A, tmp);
142
143	tmp = RREG32(VCE_CLOCK_GATING_B);
144	tmp |= 0x1e;
145	tmp &= ~0xe100e1;
146	WREG32(VCE_CLOCK_GATING_B, tmp);
147
148	tmp = RREG32(VCE_UENC_CLOCK_GATING);
149	tmp &= ~0xff9ff000;
150	WREG32(VCE_UENC_CLOCK_GATING, tmp);
151
152	tmp = RREG32(VCE_UENC_REG_CLOCK_GATING);
153	tmp &= ~0x3ff;
154	WREG32(VCE_UENC_REG_CLOCK_GATING, tmp);
155}
156
157int vce_v1_0_load_fw(struct radeon_device *rdev, uint32_t *data)
158{
159	struct vce_v1_0_fw_signature *sign = (void*)rdev->vce_fw->data;
160	uint32_t chip_id;
161	int i;
162
163	switch (rdev->family) {
164	case CHIP_TAHITI:
165		chip_id = 0x01000014;
166		break;
167	case CHIP_VERDE:
168		chip_id = 0x01000015;
169		break;
170	case CHIP_PITCAIRN:
171		chip_id = 0x01000016;
172		break;
173	case CHIP_ARUBA:
174		chip_id = 0x01000017;
175		break;
176	default:
177		return -EINVAL;
178	}
179
180	for (i = 0; i < le32_to_cpu(sign->num); ++i) {
181		if (le32_to_cpu(sign->val[i].chip_id) == chip_id)
182			break;
183	}
184
185	if (i == le32_to_cpu(sign->num))
186		return -EINVAL;
187
188	data += (256 - 64) / 4;
189	data[0] = sign->val[i].nonce[0];
190	data[1] = sign->val[i].nonce[1];
191	data[2] = sign->val[i].nonce[2];
192	data[3] = sign->val[i].nonce[3];
193	data[4] = cpu_to_le32(le32_to_cpu(sign->len) + 64);
194
195	memset(&data[5], 0, 44);
196	memcpy(&data[16], &sign[1], rdev->vce_fw->size - sizeof(*sign));
197
198	data += (le32_to_cpu(sign->len) + 64) / 4;
199	data[0] = sign->val[i].sigval[0];
200	data[1] = sign->val[i].sigval[1];
201	data[2] = sign->val[i].sigval[2];
202	data[3] = sign->val[i].sigval[3];
203
204	rdev->vce.keyselect = le32_to_cpu(sign->val[i].keyselect);
205
206	return 0;
207}
208
209unsigned vce_v1_0_bo_size(struct radeon_device *rdev)
210{
211	WARN_ON(VCE_V1_0_FW_SIZE < rdev->vce_fw->size);
212	return VCE_V1_0_FW_SIZE + VCE_V1_0_STACK_SIZE + VCE_V1_0_DATA_SIZE;
213}
214
215int vce_v1_0_resume(struct radeon_device *rdev)
216{
217	uint64_t addr = rdev->vce.gpu_addr;
218	uint32_t size;
219	int i;
220
221	WREG32_P(VCE_CLOCK_GATING_A, 0, ~(1 << 16));
222	WREG32_P(VCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000);
223	WREG32_P(VCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F);
224	WREG32(VCE_CLOCK_GATING_B, 0);
225
226	WREG32_P(VCE_LMI_FW_PERIODIC_CTRL, 0x4, ~0x4);
227
228	WREG32(VCE_LMI_CTRL, 0x00398000);
229	WREG32_P(VCE_LMI_CACHE_CTRL, 0x0, ~0x1);
230	WREG32(VCE_LMI_SWAP_CNTL, 0);
231	WREG32(VCE_LMI_SWAP_CNTL1, 0);
232	WREG32(VCE_LMI_VM_CTRL, 0);
233
234	WREG32(VCE_VCPU_SCRATCH7, RADEON_MAX_VCE_HANDLES);
235
236	addr += 256;
237	size = VCE_V1_0_FW_SIZE;
238	WREG32(VCE_VCPU_CACHE_OFFSET0, addr & 0x7fffffff);
239	WREG32(VCE_VCPU_CACHE_SIZE0, size);
240
241	addr += size;
242	size = VCE_V1_0_STACK_SIZE;
243	WREG32(VCE_VCPU_CACHE_OFFSET1, addr & 0x7fffffff);
244	WREG32(VCE_VCPU_CACHE_SIZE1, size);
245
246	addr += size;
247	size = VCE_V1_0_DATA_SIZE;
248	WREG32(VCE_VCPU_CACHE_OFFSET2, addr & 0x7fffffff);
249	WREG32(VCE_VCPU_CACHE_SIZE2, size);
250
251	WREG32_P(VCE_LMI_CTRL2, 0x0, ~0x100);
252
253	WREG32(VCE_LMI_FW_START_KEYSEL, rdev->vce.keyselect);
254
255	for (i = 0; i < 10; ++i) {
256		mdelay(10);
257		if (RREG32(VCE_FW_REG_STATUS) & VCE_FW_REG_STATUS_DONE)
258			break;
259	}
260
261	if (i == 10)
262		return -ETIMEDOUT;
263
264	if (!(RREG32(VCE_FW_REG_STATUS) & VCE_FW_REG_STATUS_PASS))
265		return -EINVAL;
266
267	for (i = 0; i < 10; ++i) {
268		mdelay(10);
269		if (!(RREG32(VCE_FW_REG_STATUS) & VCE_FW_REG_STATUS_BUSY))
270			break;
271	}
272
273	if (i == 10)
274		return -ETIMEDOUT;
275
276	vce_v1_0_init_cg(rdev);
277
278	return 0;
279}
280
281/**
282 * vce_v1_0_start - start VCE block
283 *
284 * @rdev: radeon_device pointer
285 *
286 * Setup and start the VCE block
287 */
288int vce_v1_0_start(struct radeon_device *rdev)
289{
290	struct radeon_ring *ring;
291	int i, j, r;
292
293	/* set BUSY flag */
294	WREG32_P(VCE_STATUS, 1, ~1);
295
296	ring = &rdev->ring[TN_RING_TYPE_VCE1_INDEX];
297	WREG32(VCE_RB_RPTR, ring->wptr);
298	WREG32(VCE_RB_WPTR, ring->wptr);
299	WREG32(VCE_RB_BASE_LO, ring->gpu_addr);
300	WREG32(VCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
301	WREG32(VCE_RB_SIZE, ring->ring_size / 4);
302
303	ring = &rdev->ring[TN_RING_TYPE_VCE2_INDEX];
304	WREG32(VCE_RB_RPTR2, ring->wptr);
305	WREG32(VCE_RB_WPTR2, ring->wptr);
306	WREG32(VCE_RB_BASE_LO2, ring->gpu_addr);
307	WREG32(VCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
308	WREG32(VCE_RB_SIZE2, ring->ring_size / 4);
309
310	WREG32_P(VCE_VCPU_CNTL, VCE_CLK_EN, ~VCE_CLK_EN);
311
312	WREG32_P(VCE_SOFT_RESET,
313		 VCE_ECPU_SOFT_RESET |
314		 VCE_FME_SOFT_RESET, ~(
315		 VCE_ECPU_SOFT_RESET |
316		 VCE_FME_SOFT_RESET));
317
318	mdelay(100);
319
320	WREG32_P(VCE_SOFT_RESET, 0, ~(
321		 VCE_ECPU_SOFT_RESET |
322		 VCE_FME_SOFT_RESET));
323
324	for (i = 0; i < 10; ++i) {
325		uint32_t status;
326		for (j = 0; j < 100; ++j) {
327			status = RREG32(VCE_STATUS);
328			if (status & 2)
329				break;
330			mdelay(10);
331		}
332		r = 0;
333		if (status & 2)
334			break;
335
336		DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n");
337		WREG32_P(VCE_SOFT_RESET, VCE_ECPU_SOFT_RESET, ~VCE_ECPU_SOFT_RESET);
338		mdelay(10);
339		WREG32_P(VCE_SOFT_RESET, 0, ~VCE_ECPU_SOFT_RESET);
340		mdelay(10);
341		r = -1;
342	}
343
344	/* clear BUSY flag */
345	WREG32_P(VCE_STATUS, 0, ~1);
346
347	if (r) {
348		DRM_ERROR("VCE not responding, giving up!!!\n");
349		return r;
350	}
351
352	return 0;
353}
354
355int vce_v1_0_init(struct radeon_device *rdev)
356{
357	struct radeon_ring *ring;
358	int r;
359
360	r = vce_v1_0_start(rdev);
361	if (r)
362		return r;
363
364	ring = &rdev->ring[TN_RING_TYPE_VCE1_INDEX];
365	ring->ready = true;
366	r = radeon_ring_test(rdev, TN_RING_TYPE_VCE1_INDEX, ring);
367	if (r) {
368		ring->ready = false;
369		return r;
370	}
371
372	ring = &rdev->ring[TN_RING_TYPE_VCE2_INDEX];
373	ring->ready = true;
374	r = radeon_ring_test(rdev, TN_RING_TYPE_VCE2_INDEX, ring);
375	if (r) {
376		ring->ready = false;
377		return r;
378	}
379
380	DRM_INFO("VCE initialized successfully.\n");
381
382	return 0;
383}
384