summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c1029
1 files changed, 417 insertions, 612 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
index 38f787a560cb..313fc1b53999 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
@@ -22,6 +22,7 @@
*/
#include <linux/firmware.h>
+#include <drm/drm_drv.h>
#include "amdgpu.h"
#include "amdgpu_vcn.h"
@@ -29,6 +30,8 @@
#include "soc15d.h"
#include "amdgpu_pm.h"
#include "amdgpu_psp.h"
+#include "mmsch_v2_0.h"
+#include "vcn_v2_0.h"
#include "vcn/vcn_2_0_0_offset.h"
#include "vcn/vcn_2_0_0_sh_mask.h"
@@ -43,44 +46,18 @@
#define mmUVD_SCRATCH9_INTERNAL_OFFSET 0xc01d
#define mmUVD_LMI_RBC_IB_VMID_INTERNAL_OFFSET 0x1e1
-#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x5a6
+#define mmUVD_LMI_RBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x5a6
#define mmUVD_LMI_RBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x5a7
#define mmUVD_RBC_IB_SIZE_INTERNAL_OFFSET 0x1e2
-#define mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET 0x1bfff
-#define mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET 0x4029
-#define mmUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET 0x402a
-#define mmUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET 0x402b
-#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40ea
-#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40eb
-#define mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET 0x40cf
-#define mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET 0x40d1
-#define mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40e8
-#define mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40e9
-#define mmUVD_JRBC_IB_SIZE_INTERNAL_OFFSET 0x4082
-#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40ec
-#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40ed
-#define mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET 0x4085
-#define mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET 0x4084
-#define mmUVD_JRBC_STATUS_INTERNAL_OFFSET 0x4089
-#define mmUVD_JPEG_PITCH_INTERNAL_OFFSET 0x401f
-
-#define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR 0x18000
-
-#define mmUVD_RBC_XX_IB_REG_CHECK 0x026b
-#define mmUVD_RBC_XX_IB_REG_CHECK_BASE_IDX 1
-#define mmUVD_REG_XX_MASK 0x026c
-#define mmUVD_REG_XX_MASK_BASE_IDX 1
-
static void vcn_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev);
static void vcn_v2_0_set_enc_ring_funcs(struct amdgpu_device *adev);
-static void vcn_v2_0_set_jpeg_ring_funcs(struct amdgpu_device *adev);
static void vcn_v2_0_set_irq_funcs(struct amdgpu_device *adev);
static int vcn_v2_0_set_powergating_state(void *handle,
enum amd_powergating_state state);
static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
- struct dpg_pause_state *new_state);
-
+ int inst_idx, struct dpg_pause_state *new_state);
+static int vcn_v2_0_start_sriov(struct amdgpu_device *adev);
/**
* vcn_v2_0_early_init - set function pointers
*
@@ -92,12 +69,13 @@ static int vcn_v2_0_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- adev->vcn.num_vcn_inst = 1;
- adev->vcn.num_enc_rings = 2;
+ if (amdgpu_sriov_vf(adev))
+ adev->vcn.num_enc_rings = 1;
+ else
+ adev->vcn.num_enc_rings = 2;
vcn_v2_0_set_dec_ring_funcs(adev);
vcn_v2_0_set_enc_ring_funcs(adev);
- vcn_v2_0_set_jpeg_ring_funcs(adev);
vcn_v2_0_set_irq_funcs(adev);
return 0;
@@ -115,6 +93,7 @@ static int vcn_v2_0_sw_init(void *handle)
struct amdgpu_ring *ring;
int i, r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ volatile struct amdgpu_fw_shared *fw_shared;
/* VCN DEC TRAP */
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
@@ -132,25 +111,11 @@ static int vcn_v2_0_sw_init(void *handle)
return r;
}
- /* VCN JPEG TRAP */
- r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN,
- VCN_2_0__SRCID__JPEG_DECODE, &adev->vcn.inst->irq);
- if (r)
- return r;
-
r = amdgpu_vcn_sw_init(adev);
if (r)
return r;
- if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
- const struct common_firmware_header *hdr;
- hdr = (const struct common_firmware_header *)adev->vcn.fw->data;
- adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].ucode_id = AMDGPU_UCODE_ID_VCN;
- adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].fw = adev->vcn.fw;
- adev->firmware.fw_size +=
- ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE);
- DRM_INFO("PSP loading VCN firmware\n");
- }
+ amdgpu_vcn_setup_ucode(adev);
r = amdgpu_vcn_resume(adev);
if (r)
@@ -162,7 +127,8 @@ static int vcn_v2_0_sw_init(void *handle)
ring->doorbell_index = adev->doorbell_index.vcn.vcn_ring0_1 << 1;
sprintf(ring->name, "vcn_dec");
- r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0);
+ r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
+ AMDGPU_RING_PRIO_DEFAULT, NULL);
if (r)
return r;
@@ -185,28 +151,29 @@ static int vcn_v2_0_sw_init(void *handle)
adev->vcn.inst->external.nop = SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP);
for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+ enum amdgpu_ring_priority_level hw_prio = amdgpu_vcn_get_enc_ring_prio(i);
+
ring = &adev->vcn.inst->ring_enc[i];
ring->use_doorbell = true;
- ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + i;
+ if (!amdgpu_sriov_vf(adev))
+ ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 2 + i;
+ else
+ ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1 + i;
sprintf(ring->name, "vcn_enc%d", i);
- r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0);
+ r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0,
+ hw_prio, NULL);
if (r)
return r;
}
- ring = &adev->vcn.inst->ring_jpeg;
- ring->use_doorbell = true;
- ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1;
- sprintf(ring->name, "vcn_jpeg");
- r = amdgpu_ring_init(adev, ring, 512, &adev->vcn.inst->irq, 0);
- if (r)
- return r;
-
adev->vcn.pause_dpg_mode = vcn_v2_0_pause_dpg_mode;
- adev->vcn.internal.jpeg_pitch = mmUVD_JPEG_PITCH_INTERNAL_OFFSET;
- adev->vcn.inst->external.jpeg_pitch = SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_PITCH);
+ r = amdgpu_virt_alloc_mm_table(adev);
+ if (r)
+ return r;
+ fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
+ fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_VCN_MULTI_QUEUE_FLAG);
return 0;
}
@@ -219,8 +186,16 @@ static int vcn_v2_0_sw_init(void *handle)
*/
static int vcn_v2_0_sw_fini(void *handle)
{
- int r;
+ int r, idx;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
+
+ if (drm_dev_enter(adev_to_drm(adev), &idx)) {
+ fw_shared->present_flag_0 = 0;
+ drm_dev_exit(idx);
+ }
+
+ amdgpu_virt_free_mm_table(adev);
r = amdgpu_vcn_suspend(adev);
if (r)
@@ -247,10 +222,17 @@ static int vcn_v2_0_hw_init(void *handle)
adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell,
ring->doorbell_index, 0);
+ if (amdgpu_sriov_vf(adev))
+ vcn_v2_0_start_sriov(adev);
+
r = amdgpu_ring_test_helper(ring);
if (r)
goto done;
+ //Disable vcn decode for sriov
+ if (amdgpu_sriov_vf(adev))
+ ring->sched.ready = false;
+
for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
ring = &adev->vcn.inst->ring_enc[i];
r = amdgpu_ring_test_helper(ring);
@@ -258,11 +240,6 @@ static int vcn_v2_0_hw_init(void *handle)
goto done;
}
- ring = &adev->vcn.inst->ring_jpeg;
- r = amdgpu_ring_test_helper(ring);
- if (r)
- goto done;
-
done:
if (!r)
DRM_INFO("VCN decode and encode initialized successfully(under %s).\n",
@@ -281,24 +258,14 @@ done:
static int vcn_v2_0_hw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
- int i;
+
+ cancel_delayed_work_sync(&adev->vcn.idle_work);
if ((adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) ||
(adev->vcn.cur_state != AMD_PG_STATE_GATE &&
RREG32_SOC15(VCN, 0, mmUVD_STATUS)))
vcn_v2_0_set_powergating_state(adev, AMD_PG_STATE_GATE);
- ring->sched.ready = false;
-
- for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
- ring = &adev->vcn.inst->ring_enc[i];
- ring->sched.ready = false;
- }
-
- ring = &adev->vcn.inst->ring_jpeg;
- ring->sched.ready = false;
-
return 0;
}
@@ -356,6 +323,9 @@ static void vcn_v2_0_mc_resume(struct amdgpu_device *adev)
uint32_t size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
uint32_t offset;
+ if (amdgpu_sriov_vf(adev))
+ return;
+
/* cache window 0: fw */
if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW,
@@ -392,8 +362,16 @@ static void vcn_v2_0_mc_resume(struct amdgpu_device *adev)
WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_OFFSET2, 0);
WREG32_SOC15(UVD, 0, mmUVD_VCPU_CACHE_SIZE2, AMDGPU_VCN_CONTEXT_SIZE);
+ /* non-cache window */
+ WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW,
+ lower_32_bits(adev->vcn.inst->fw_shared_gpu_addr));
+ WREG32_SOC15(UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH,
+ upper_32_bits(adev->vcn.inst->fw_shared_gpu_addr));
+ WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0, 0);
+ WREG32_SOC15(UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0,
+ AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)));
+
WREG32_SOC15(UVD, 0, mmUVD_GFX10_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
- WREG32_SOC15(UVD, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG, adev->gfx.config.gb_addr_config);
}
static void vcn_v2_0_mc_resume_dpg_mode(struct amdgpu_device *adev, bool indirect)
@@ -404,88 +382,91 @@ static void vcn_v2_0_mc_resume_dpg_mode(struct amdgpu_device *adev, bool indirec
/* cache window 0: fw */
if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
if (!indirect) {
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
(adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
} else {
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_OFFSET0), 0, 0, indirect);
}
offset = 0;
} else {
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
lower_32_bits(adev->vcn.inst->gpu_addr), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
upper_32_bits(adev->vcn.inst->gpu_addr), 0, indirect);
offset = size;
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_OFFSET0),
AMDGPU_UVD_FIRMWARE_OFFSET >> 3, 0, indirect);
}
if (!indirect)
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_SIZE0), size, 0, indirect);
else
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_SIZE0), 0, 0, indirect);
/* cache window 1: stack */
if (!indirect) {
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
lower_32_bits(adev->vcn.inst->gpu_addr + offset), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
upper_32_bits(adev->vcn.inst->gpu_addr + offset), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
} else {
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_OFFSET1), 0, 0, indirect);
}
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_SIZE1), AMDGPU_VCN_STACK_SIZE, 0, indirect);
/* cache window 2: context */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
lower_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
upper_32_bits(adev->vcn.inst->gpu_addr + offset + AMDGPU_VCN_STACK_SIZE), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_OFFSET2), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CACHE_SIZE2), AMDGPU_VCN_CONTEXT_SIZE, 0, indirect);
/* non-cache window */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
- UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
- UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+ UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst->fw_shared_gpu_addr), 0, indirect);
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+ UVD, 0, mmUVD_LMI_VCPU_NC0_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst->fw_shared_gpu_addr), 0, indirect);
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_NONCACHE_OFFSET0), 0, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
- UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0), 0, 0, indirect);
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
+ UVD, 0, mmUVD_VCPU_NONCACHE_SIZE0),
+ AMDGPU_GPU_PAGE_ALIGN(sizeof(struct amdgpu_fw_shared)), 0, indirect);
/* VCN global tiling registers */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect);
}
@@ -493,7 +474,6 @@ static void vcn_v2_0_mc_resume_dpg_mode(struct amdgpu_device *adev, bool indirec
* vcn_v2_0_disable_clock_gating - disable VCN clock gating
*
* @adev: amdgpu_device pointer
- * @sw: enable SW clock gating
*
* Disable clock gating for VCN block
*/
@@ -501,6 +481,9 @@ static void vcn_v2_0_disable_clock_gating(struct amdgpu_device *adev)
{
uint32_t data;
+ if (amdgpu_sriov_vf(adev))
+ return;
+
/* UVD disable CGC */
data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
@@ -631,150 +614,26 @@ static void vcn_v2_0_clock_gating_dpg_mode(struct amdgpu_device *adev,
UVD_CGC_CTRL__WCB_MODE_MASK |
UVD_CGC_CTRL__VCPU_MODE_MASK |
UVD_CGC_CTRL__SCPU_MODE_MASK);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_CGC_CTRL), reg_data, sram_sel, indirect);
/* turn off clock gating */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_CGC_GATE), 0, sram_sel, indirect);
/* turn on SUVD clock gating */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_SUVD_CGC_GATE), 1, sram_sel, indirect);
/* turn on sw mode in UVD_SUVD_CGC_CTRL */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_SUVD_CGC_CTRL), 0, sram_sel, indirect);
}
/**
- * jpeg_v2_0_start - start JPEG block
- *
- * @adev: amdgpu_device pointer
- *
- * Setup and start the JPEG block
- */
-static int jpeg_v2_0_start(struct amdgpu_device *adev)
-{
- struct amdgpu_ring *ring = &adev->vcn.inst->ring_jpeg;
- uint32_t tmp;
- int r = 0;
-
- /* disable power gating */
- tmp = 1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
- WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_PGFSM_CONFIG), tmp);
-
- SOC15_WAIT_ON_RREG(VCN, 0,
- mmUVD_PGFSM_STATUS, UVD_PGFSM_STATUS_UVDJ_PWR_ON,
- UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK, r);
-
- if (r) {
- DRM_ERROR("amdgpu: JPEG disable power gating failed\n");
- return r;
- }
-
- /* Removing the anti hang mechanism to indicate the UVDJ tile is ON */
- tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_POWER_STATUS)) & ~0x1;
- WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_POWER_STATUS), tmp);
-
- /* JPEG disable CGC */
- tmp = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL);
- tmp |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
- tmp |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
- tmp |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
- WREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL, tmp);
-
- tmp = RREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE);
- tmp &= ~(JPEG_CGC_GATE__JPEG_DEC_MASK
- | JPEG_CGC_GATE__JPEG2_DEC_MASK
- | JPEG_CGC_GATE__JPEG_ENC_MASK
- | JPEG_CGC_GATE__JMCIF_MASK
- | JPEG_CGC_GATE__JRBBM_MASK);
- WREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE, tmp);
-
- /* enable JMI channel */
- WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_JMI_CNTL), 0,
- ~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
- /* enable System Interrupt for JRBC */
- WREG32_P(SOC15_REG_OFFSET(VCN, 0, mmJPEG_SYS_INT_EN),
- JPEG_SYS_INT_EN__DJRBC_MASK,
- ~JPEG_SYS_INT_EN__DJRBC_MASK);
-
- WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_VMID, 0);
- WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, (0x00000001L | 0x00000002L));
- WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_LOW,
- lower_32_bits(ring->gpu_addr));
- WREG32_SOC15(UVD, 0, mmUVD_LMI_JRBC_RB_64BIT_BAR_HIGH,
- upper_32_bits(ring->gpu_addr));
- WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR, 0);
- WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, 0);
- WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_CNTL, 0x00000002L);
- WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_SIZE, ring->ring_size / 4);
- ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
-
- return 0;
-}
-
-/**
- * jpeg_v2_0_stop - stop JPEG block
- *
- * @adev: amdgpu_device pointer
- *
- * stop the JPEG block
- */
-static int jpeg_v2_0_stop(struct amdgpu_device *adev)
-{
- uint32_t tmp;
- int r = 0;
-
- /* reset JMI */
- WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_JMI_CNTL),
- UVD_JMI_CNTL__SOFT_RESET_MASK,
- ~UVD_JMI_CNTL__SOFT_RESET_MASK);
-
- /* enable JPEG CGC */
- tmp = RREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL);
- tmp |= 1 << JPEG_CGC_CTRL__DYN_CLOCK_MODE__SHIFT;
- tmp |= 1 << JPEG_CGC_CTRL__CLK_GATE_DLY_TIMER__SHIFT;
- tmp |= 4 << JPEG_CGC_CTRL__CLK_OFF_DELAY__SHIFT;
- WREG32_SOC15(VCN, 0, mmJPEG_CGC_CTRL, tmp);
-
-
- tmp = RREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE);
- tmp |= (JPEG_CGC_GATE__JPEG_DEC_MASK
- |JPEG_CGC_GATE__JPEG2_DEC_MASK
- |JPEG_CGC_GATE__JPEG_ENC_MASK
- |JPEG_CGC_GATE__JMCIF_MASK
- |JPEG_CGC_GATE__JRBBM_MASK);
- WREG32_SOC15(VCN, 0, mmJPEG_CGC_GATE, tmp);
-
- /* enable power gating */
- tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_POWER_STATUS));
- tmp &= ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK;
- tmp |= 0x1; //UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_TILES_OFF;
- WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_POWER_STATUS), tmp);
-
- tmp = 2 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT;
- WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_PGFSM_CONFIG), tmp);
-
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS,
- (2 << UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT),
- UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK, r);
-
- if (r) {
- DRM_ERROR("amdgpu: JPEG enable power gating failed\n");
- return r;
- }
-
- return r;
-}
-
-/**
* vcn_v2_0_enable_clock_gating - enable VCN clock gating
*
* @adev: amdgpu_device pointer
- * @sw: enable SW clock gating
*
* Enable clock gating for VCN block
*/
@@ -782,6 +641,9 @@ static void vcn_v2_0_enable_clock_gating(struct amdgpu_device *adev)
{
uint32_t data = 0;
+ if (amdgpu_sriov_vf(adev))
+ return;
+
/* enable UVD CGC */
data = RREG32_SOC15(VCN, 0, mmUVD_CGC_CTRL);
if (adev->cg_flags & AMD_CG_SUPPORT_VCN_MGCG)
@@ -832,7 +694,9 @@ static void vcn_v2_0_enable_clock_gating(struct amdgpu_device *adev)
static void vcn_v2_0_disable_static_power_gating(struct amdgpu_device *adev)
{
uint32_t data = 0;
- int ret;
+
+ if (amdgpu_sriov_vf(adev))
+ return;
if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
@@ -848,7 +712,7 @@ static void vcn_v2_0_disable_static_power_gating(struct amdgpu_device *adev)
WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS,
- UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON_2_0, 0xFFFFF, ret);
+ UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON_2_0, 0xFFFFF);
} else {
data = (1 << UVD_PGFSM_CONFIG__UVDM_PWR_CONFIG__SHIFT
| 1 << UVD_PGFSM_CONFIG__UVDU_PWR_CONFIG__SHIFT
@@ -861,7 +725,7 @@ static void vcn_v2_0_disable_static_power_gating(struct amdgpu_device *adev)
| 1 << UVD_PGFSM_CONFIG__UVDTE_PWR_CONFIG__SHIFT
| 1 << UVD_PGFSM_CONFIG__UVDE_PWR_CONFIG__SHIFT);
WREG32_SOC15(VCN, 0, mmUVD_PGFSM_CONFIG, data);
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, 0, 0xFFFFF, ret);
+ SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, 0, 0xFFFFF);
}
/* polling UVD_PGFSM_STATUS to confirm UVDM_PWR_STATUS,
@@ -879,7 +743,9 @@ static void vcn_v2_0_disable_static_power_gating(struct amdgpu_device *adev)
static void vcn_v2_0_enable_static_power_gating(struct amdgpu_device *adev)
{
uint32_t data = 0;
- int ret;
+
+ if (amdgpu_sriov_vf(adev))
+ return;
if (adev->pg_flags & AMD_PG_SUPPORT_VCN) {
/* Before power off, this indicator has to be turned on */
@@ -912,12 +778,13 @@ static void vcn_v2_0_enable_static_power_gating(struct amdgpu_device *adev)
| 2 << UVD_PGFSM_STATUS__UVDTD_PWR_STATUS__SHIFT
| 2 << UVD_PGFSM_STATUS__UVDTE_PWR_STATUS__SHIFT
| 2 << UVD_PGFSM_STATUS__UVDE_PWR_STATUS__SHIFT);
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, data, 0xFFFFF, ret);
+ SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_PGFSM_STATUS, data, 0xFFFFF);
}
}
static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
{
+ volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
uint32_t rb_bufsz, tmp;
@@ -930,7 +797,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
WREG32_SOC15(UVD, 0, mmUVD_POWER_STATUS, tmp);
if (indirect)
- adev->vcn.dpg_sram_curr_addr = (uint32_t*)adev->vcn.dpg_sram_cpu_addr;
+ adev->vcn.inst->dpg_sram_curr_addr = (uint32_t *)adev->vcn.inst->dpg_sram_cpu_addr;
/* enable clock gating */
vcn_v2_0_clock_gating_dpg_mode(adev, 0, indirect);
@@ -939,11 +806,11 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
tmp = (0xFF << UVD_VCPU_CNTL__PRB_TIMEOUT_VAL__SHIFT);
tmp |= UVD_VCPU_CNTL__CLK_EN_MASK;
tmp |= UVD_VCPU_CNTL__MIF_WR_LOW_THRESHOLD_BP_MASK;
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_VCPU_CNTL), tmp, 0, indirect);
/* disable master interupt */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_MASTINT_EN), 0, 0, indirect);
/* setup mmUVD_LMI_CTRL */
@@ -955,28 +822,28 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK |
(8 << UVD_LMI_CTRL__WRITE_CLEAN_TIMER__SHIFT) |
0x00100000L);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_CTRL), tmp, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_MPC_CNTL),
0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_MPC_SET_MUXA0),
((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) |
(0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) |
(0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) |
(0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT)), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_MPC_SET_MUXB0),
((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) |
(0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) |
(0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) |
(0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT)), 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_MPC_SET_MUX),
((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) |
(0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) |
@@ -984,29 +851,29 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
vcn_v2_0_mc_resume_dpg_mode(adev, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_REG_XX_MASK), 0x10, 0, indirect);
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_RBC_XX_IB_REG_CHECK), 0x3, 0, indirect);
/* release VCPU reset to boot */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_SOFT_RESET), 0, 0, indirect);
/* enable LMI MC and UMC channels */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_LMI_CTRL2),
0x1F << UVD_LMI_CTRL2__RE_OFLD_MIF_WR_REQ_NUM__SHIFT, 0, indirect);
/* enable master interrupt */
- WREG32_SOC15_DPG_MODE_2_0(SOC15_DPG_MODE_OFFSET_2_0(
+ WREG32_SOC15_DPG_MODE(0, SOC15_DPG_MODE_OFFSET(
UVD, 0, mmUVD_MASTINT_EN),
UVD_MASTINT_EN__VCPU_EN_MASK, 0, indirect);
if (indirect)
- psp_update_vcn_sram(adev, 0, adev->vcn.dpg_sram_gpu_addr,
- (uint32_t)((uintptr_t)adev->vcn.dpg_sram_curr_addr -
- (uintptr_t)adev->vcn.dpg_sram_cpu_addr));
+ psp_update_vcn_sram(adev, 0, adev->vcn.inst->dpg_sram_gpu_addr,
+ (uint32_t)((uintptr_t)adev->vcn.inst->dpg_sram_curr_addr -
+ (uintptr_t)adev->vcn.inst->dpg_sram_cpu_addr));
/* force RBC into idle state */
rb_bufsz = order_base_2(ring->ring_size);
@@ -1017,6 +884,12 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
+ /* Stall DPG before WPTR/RPTR reset */
+ WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+ UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+ ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
+ fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+
/* set the write pointer delay */
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL, 0);
@@ -1024,7 +897,7 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR,
(upper_32_bits(ring->gpu_addr) >> 2));
- /* programm the RB_BASE for ring buffer */
+ /* program the RB_BASE for ring buffer */
WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
lower_32_bits(ring->gpu_addr));
WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
@@ -1039,11 +912,16 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_device *adev, bool indirect)
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
lower_32_bits(ring->wptr));
+ fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+ /* Unstall DPG */
+ WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+ 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
return 0;
}
static int vcn_v2_0_start(struct amdgpu_device *adev)
{
+ volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
struct amdgpu_ring *ring = &adev->vcn.inst->ring_dec;
uint32_t rb_bufsz, tmp;
uint32_t lmi_swap_cntl;
@@ -1052,12 +930,8 @@ static int vcn_v2_0_start(struct amdgpu_device *adev)
if (adev->pm.dpm_enabled)
amdgpu_dpm_enable_uvd(adev, true);
- if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
- r = vcn_v2_0_start_dpg_mode(adev, adev->vcn.indirect_sram);
- if (r)
- return r;
- goto jpeg;
- }
+ if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG)
+ return vcn_v2_0_start_dpg_mode(adev, adev->vcn.indirect_sram);
vcn_v2_0_disable_static_power_gating(adev);
@@ -1182,7 +1056,8 @@ static int vcn_v2_0_start(struct amdgpu_device *adev)
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_CNTL, tmp);
- /* programm the RB_BASE for ring buffer */
+ fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
+ /* program the RB_BASE for ring buffer */
WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW,
lower_32_bits(ring->gpu_addr));
WREG32_SOC15(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH,
@@ -1194,51 +1069,49 @@ static int vcn_v2_0_start(struct amdgpu_device *adev)
ring->wptr = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_RPTR);
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
lower_32_bits(ring->wptr));
+ fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+ fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
ring = &adev->vcn.inst->ring_enc[0];
WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
+ fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
+ fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
ring = &adev->vcn.inst->ring_enc[1];
WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
+ fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
-jpeg:
- r = jpeg_v2_0_start(adev);
-
- return r;
+ return 0;
}
static int vcn_v2_0_stop_dpg_mode(struct amdgpu_device *adev)
{
- int ret_code = 0;
uint32_t tmp;
/* Wait for power status to be 1 */
SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 1,
- UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+ UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
/* wait for read ptr to be equal to write ptr */
tmp = RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR);
- SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR, tmp, 0xFFFFFFFF, ret_code);
+ SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR, tmp, 0xFFFFFFFF);
tmp = RREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2);
- SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR2, tmp, 0xFFFFFFFF, ret_code);
-
- tmp = RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
- SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_JRBC_RB_RPTR, tmp, 0xFFFFFFFF, ret_code);
+ SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RB_RPTR2, tmp, 0xFFFFFFFF);
tmp = RREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR) & 0x7FFFFFFF;
- SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RBC_RB_RPTR, tmp, 0xFFFFFFFF, ret_code);
+ SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_RBC_RB_RPTR, tmp, 0xFFFFFFFF);
SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 1,
- UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+ UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
/* disable dynamic power gating mode */
WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0,
@@ -1252,10 +1125,6 @@ static int vcn_v2_0_stop(struct amdgpu_device *adev)
uint32_t tmp;
int r;
- r = jpeg_v2_0_stop(adev);
- if (r)
- return r;
-
if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) {
r = vcn_v2_0_stop_dpg_mode(adev);
if (r)
@@ -1264,7 +1133,7 @@ static int vcn_v2_0_stop(struct amdgpu_device *adev)
}
/* wait for uvd idle */
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE, 0x7, r);
+ r = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE, 0x7);
if (r)
return r;
@@ -1272,7 +1141,7 @@ static int vcn_v2_0_stop(struct amdgpu_device *adev)
UVD_LMI_STATUS__READ_CLEAN_MASK |
UVD_LMI_STATUS__WRITE_CLEAN_MASK |
UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK;
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_LMI_STATUS, tmp, tmp, r);
+ r = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_LMI_STATUS, tmp, tmp);
if (r)
return r;
@@ -1283,7 +1152,7 @@ static int vcn_v2_0_stop(struct amdgpu_device *adev)
tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK|
UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK;
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_LMI_STATUS, tmp, tmp, r);
+ r = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_LMI_STATUS, tmp, tmp);
if (r)
return r;
@@ -1320,25 +1189,25 @@ power_off:
}
static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
- struct dpg_pause_state *new_state)
+ int inst_idx, struct dpg_pause_state *new_state)
{
struct amdgpu_ring *ring;
uint32_t reg_data = 0;
int ret_code;
/* pause/unpause if state is changed */
- if (adev->vcn.pause_state.fw_based != new_state->fw_based) {
+ if (adev->vcn.inst[inst_idx].pause_state.fw_based != new_state->fw_based) {
DRM_DEBUG("dpg pause state changed %d -> %d",
- adev->vcn.pause_state.fw_based, new_state->fw_based);
+ adev->vcn.inst[inst_idx].pause_state.fw_based, new_state->fw_based);
reg_data = RREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE) &
(~UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
if (new_state->fw_based == VCN_DPG_STATE__PAUSE) {
- ret_code = 0;
- SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 0x1,
- UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+ ret_code = SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS, 0x1,
+ UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
if (!ret_code) {
+ volatile struct amdgpu_fw_shared *fw_shared = adev->vcn.inst->fw_shared_cpu_addr;
/* pause DPG */
reg_data |= UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
@@ -1346,36 +1215,51 @@ static int vcn_v2_0_pause_dpg_mode(struct amdgpu_device *adev,
/* wait for ACK */
SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_DPG_PAUSE,
UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK,
- UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK, ret_code);
+ UVD_DPG_PAUSE__NJ_PAUSE_DPG_ACK_MASK);
+ /* Stall DPG before WPTR/RPTR reset */
+ WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+ UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK,
+ ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
/* Restore */
+ fw_shared->multi_queue.encode_generalpurpose_queue_mode |= FW_QUEUE_RING_RESET;
ring = &adev->vcn.inst->ring_enc[0];
+ ring->wptr = 0;
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO, ring->gpu_addr);
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr));
WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE, ring->ring_size / 4);
WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR, lower_32_bits(ring->wptr));
WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR, lower_32_bits(ring->wptr));
+ fw_shared->multi_queue.encode_generalpurpose_queue_mode &= ~FW_QUEUE_RING_RESET;
+ fw_shared->multi_queue.encode_lowlatency_queue_mode |= FW_QUEUE_RING_RESET;
ring = &adev->vcn.inst->ring_enc[1];
+ ring->wptr = 0;
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_LO2, ring->gpu_addr);
WREG32_SOC15(UVD, 0, mmUVD_RB_BASE_HI2, upper_32_bits(ring->gpu_addr));
WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4);
WREG32_SOC15(UVD, 0, mmUVD_RB_RPTR2, lower_32_bits(ring->wptr));
WREG32_SOC15(UVD, 0, mmUVD_RB_WPTR2, lower_32_bits(ring->wptr));
+ fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET;
+ fw_shared->multi_queue.decode_queue_mode |= FW_QUEUE_RING_RESET;
WREG32_SOC15(UVD, 0, mmUVD_RBC_RB_WPTR,
RREG32_SOC15(UVD, 0, mmUVD_SCRATCH2) & 0x7FFFFFFF);
+ fw_shared->multi_queue.decode_queue_mode &= ~FW_QUEUE_RING_RESET;
+ /* Unstall DPG */
+ WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
+ 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK);
SOC15_WAIT_ON_RREG(UVD, 0, mmUVD_POWER_STATUS,
UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON,
- UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ret_code);
+ UVD_POWER_STATUS__UVD_POWER_STATUS_MASK);
}
} else {
/* unpause dpg, no need to wait */
reg_data &= ~UVD_DPG_PAUSE__NJ_PAUSE_DPG_REQ_MASK;
WREG32_SOC15(UVD, 0, mmUVD_DPG_PAUSE, reg_data);
}
- adev->vcn.pause_state.fw_based = new_state->fw_based;
+ adev->vcn.inst[inst_idx].pause_state.fw_based = new_state->fw_based;
}
return 0;
@@ -1391,10 +1275,10 @@ static bool vcn_v2_0_is_idle(void *handle)
static int vcn_v2_0_wait_for_idle(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- int ret = 0;
+ int ret;
- SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE,
- UVD_STATUS__IDLE, ret);
+ ret = SOC15_WAIT_ON_RREG(VCN, 0, mmUVD_STATUS, UVD_STATUS__IDLE,
+ UVD_STATUS__IDLE);
return ret;
}
@@ -1403,11 +1287,14 @@ static int vcn_v2_0_set_clockgating_state(void *handle,
enum amd_clockgating_state state)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- bool enable = (state == AMD_CG_STATE_GATE) ? true : false;
+ bool enable = (state == AMD_CG_STATE_GATE);
+
+ if (amdgpu_sriov_vf(adev))
+ return 0;
if (enable) {
/* wait for STATUS to clear */
- if (vcn_v2_0_is_idle(handle))
+ if (!vcn_v2_0_is_idle(handle))
return -EBUSY;
vcn_v2_0_enable_clock_gating(adev);
} else {
@@ -1507,6 +1394,7 @@ void vcn_v2_0_dec_ring_insert_end(struct amdgpu_ring *ring)
* vcn_v2_0_dec_ring_insert_nop - insert a nop command
*
* @ring: amdgpu_ring pointer
+ * @count: the number of NOP packets to insert
*
* Write a nop command to the ring.
*/
@@ -1527,7 +1415,9 @@ void vcn_v2_0_dec_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
* vcn_v2_0_dec_ring_emit_fence - emit an fence & trap command
*
* @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
*
* Write a fence and a trap command to the ring.
*/
@@ -1564,7 +1454,9 @@ void vcn_v2_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
* vcn_v2_0_dec_ring_emit_ib - execute indirect buffer
*
* @ring: amdgpu_ring pointer
+ * @job: job to retrieve vmid from
* @ib: indirect buffer to execute
+ * @flags: unused
*
* Write ring commands to execute the indirect buffer
*/
@@ -1615,7 +1507,7 @@ void vcn_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring,
pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
/* wait for register write */
- data0 = hub->ctx0_ptb_addr_lo32 + vmid * 2;
+ data0 = hub->ctx0_ptb_addr_lo32 + vmid * hub->ctx_addr_distance;
data1 = lower_32_bits(pd_addr);
mask = 0xffffffff;
vcn_v2_0_dec_ring_emit_reg_wait(ring, data0, data1, mask);
@@ -1710,7 +1602,9 @@ static void vcn_v2_0_enc_ring_set_wptr(struct amdgpu_ring *ring)
* vcn_v2_0_enc_ring_emit_fence - emit an enc fence & trap command
*
* @ring: amdgpu_ring pointer
- * @fence: fence to emit
+ * @addr: address
+ * @seq: sequence number
+ * @flags: fence related flags
*
* Write enc a fence and a trap command to the ring.
*/
@@ -1735,7 +1629,9 @@ void vcn_v2_0_enc_ring_insert_end(struct amdgpu_ring *ring)
* vcn_v2_0_enc_ring_emit_ib - enc execute indirect buffer
*
* @ring: amdgpu_ring pointer
+ * @job: job to retrive vmid from
* @ib: indirect buffer to execute
+ * @flags: unused
*
* Write enc ring commands to execute the indirect buffer
*/
@@ -1770,7 +1666,8 @@ void vcn_v2_0_enc_ring_emit_vm_flush(struct amdgpu_ring *ring,
pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
/* wait for reg writes */
- vcn_v2_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 + vmid * 2,
+ vcn_v2_0_enc_ring_emit_reg_wait(ring, hub->ctx0_ptb_addr_lo32 +
+ vmid * hub->ctx_addr_distance,
lower_32_bits(pd_addr), 0xffffffff);
}
@@ -1781,272 +1678,6 @@ void vcn_v2_0_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_
amdgpu_ring_write(ring, val);
}
-/**
- * vcn_v2_0_jpeg_ring_get_rptr - get read pointer
- *
- * @ring: amdgpu_ring pointer
- *
- * Returns the current hardware read pointer
- */
-static uint64_t vcn_v2_0_jpeg_ring_get_rptr(struct amdgpu_ring *ring)
-{
- struct amdgpu_device *adev = ring->adev;
-
- return RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_RPTR);
-}
-
-/**
- * vcn_v2_0_jpeg_ring_get_wptr - get write pointer
- *
- * @ring: amdgpu_ring pointer
- *
- * Returns the current hardware write pointer
- */
-static uint64_t vcn_v2_0_jpeg_ring_get_wptr(struct amdgpu_ring *ring)
-{
- struct amdgpu_device *adev = ring->adev;
-
- if (ring->use_doorbell)
- return adev->wb.wb[ring->wptr_offs];
- else
- return RREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR);
-}
-
-/**
- * vcn_v2_0_jpeg_ring_set_wptr - set write pointer
- *
- * @ring: amdgpu_ring pointer
- *
- * Commits the write pointer to the hardware
- */
-static void vcn_v2_0_jpeg_ring_set_wptr(struct amdgpu_ring *ring)
-{
- struct amdgpu_device *adev = ring->adev;
-
- if (ring->use_doorbell) {
- adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr);
- WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr));
- } else {
- WREG32_SOC15(UVD, 0, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr));
- }
-}
-
-/**
- * vcn_v2_0_jpeg_ring_insert_start - insert a start command
- *
- * @ring: amdgpu_ring pointer
- *
- * Write a start command to the ring.
- */
-void vcn_v2_0_jpeg_ring_insert_start(struct amdgpu_ring *ring)
-{
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x68e04);
-
- amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x80010000);
-}
-
-/**
- * vcn_v2_0_jpeg_ring_insert_end - insert a end command
- *
- * @ring: amdgpu_ring pointer
- *
- * Write a end command to the ring.
- */
-void vcn_v2_0_jpeg_ring_insert_end(struct amdgpu_ring *ring)
-{
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x68e04);
-
- amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x00010000);
-}
-
-/**
- * vcn_v2_0_jpeg_ring_emit_fence - emit an fence & trap command
- *
- * @ring: amdgpu_ring pointer
- * @fence: fence to emit
- *
- * Write a fence and a trap command to the ring.
- */
-void vcn_v2_0_jpeg_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
- unsigned flags)
-{
- WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, seq);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, seq);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, lower_32_bits(addr));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, upper_32_bits(addr));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x8);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET,
- 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE4));
- amdgpu_ring_write(ring, 0);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x3fbc);
-
- amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x1);
-
- amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE7));
- amdgpu_ring_write(ring, 0);
-}
-
-/**
- * vcn_v2_0_jpeg_ring_emit_ib - execute indirect buffer
- *
- * @ring: amdgpu_ring pointer
- * @ib: indirect buffer to execute
- *
- * Write ring commands to execute the indirect buffer.
- */
-void vcn_v2_0_jpeg_ring_emit_ib(struct amdgpu_ring *ring,
- struct amdgpu_job *job,
- struct amdgpu_ib *ib,
- uint32_t flags)
-{
- unsigned vmid = AMDGPU_JOB_GET_VMID(job);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, (vmid | (vmid << 4)));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, (vmid | (vmid << 4)));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_IB_SIZE_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, ib->length_dw);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, lower_32_bits(ring->gpu_addr));
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, upper_32_bits(ring->gpu_addr));
-
- amdgpu_ring_write(ring, PACKETJ(0, 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE2));
- amdgpu_ring_write(ring, 0);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x01400200);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x2);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_STATUS_INTERNAL_OFFSET,
- 0, PACKETJ_CONDITION_CHECK3, PACKETJ_TYPE3));
- amdgpu_ring_write(ring, 0x2);
-}
-
-void vcn_v2_0_jpeg_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg,
- uint32_t val, uint32_t mask)
-{
- uint32_t reg_offset = (reg << 2);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, 0x01400200);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- amdgpu_ring_write(ring, val);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- if (reg_offset >= 0x10000 && reg_offset <= 0x105ff) {
- amdgpu_ring_write(ring, 0);
- amdgpu_ring_write(ring,
- PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE3));
- } else {
- amdgpu_ring_write(ring, reg_offset);
- amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
- 0, 0, PACKETJ_TYPE3));
- }
- amdgpu_ring_write(ring, mask);
-}
-
-void vcn_v2_0_jpeg_ring_emit_vm_flush(struct amdgpu_ring *ring,
- unsigned vmid, uint64_t pd_addr)
-{
- struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
- uint32_t data0, data1, mask;
-
- pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr);
-
- /* wait for register write */
- data0 = hub->ctx0_ptb_addr_lo32 + vmid * 2;
- data1 = lower_32_bits(pd_addr);
- mask = 0xffffffff;
- vcn_v2_0_jpeg_ring_emit_reg_wait(ring, data0, data1, mask);
-}
-
-void vcn_v2_0_jpeg_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val)
-{
- uint32_t reg_offset = (reg << 2);
-
- amdgpu_ring_write(ring, PACKETJ(mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET,
- 0, 0, PACKETJ_TYPE0));
- if (reg_offset >= 0x10000 && reg_offset <= 0x105ff) {
- amdgpu_ring_write(ring, 0);
- amdgpu_ring_write(ring,
- PACKETJ((reg_offset >> 2), 0, 0, PACKETJ_TYPE0));
- } else {
- amdgpu_ring_write(ring, reg_offset);
- amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR,
- 0, 0, PACKETJ_TYPE0));
- }
- amdgpu_ring_write(ring, val);
-}
-
-void vcn_v2_0_jpeg_ring_nop(struct amdgpu_ring *ring, uint32_t count)
-{
- int i;
-
- WARN_ON(ring->wptr % 2 || count % 2);
-
- for (i = 0; i < count / 2; i++) {
- amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6));
- amdgpu_ring_write(ring, 0);
- }
-}
-
static int vcn_v2_0_set_interrupt_state(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
unsigned type,
@@ -2071,9 +1702,6 @@ static int vcn_v2_0_process_interrupt(struct amdgpu_device *adev,
case VCN_2_0__SRCID__UVD_ENC_LOW_LATENCY:
amdgpu_fence_process(&adev->vcn.inst->ring_enc[1]);
break;
- case VCN_2_0__SRCID__JPEG_DECODE:
- amdgpu_fence_process(&adev->vcn.inst->ring_jpeg);
- break;
default:
DRM_ERROR("Unhandled interrupt: %d %d\n",
entry->src_id, entry->src_data[0]);
@@ -2083,13 +1711,16 @@ static int vcn_v2_0_process_interrupt(struct amdgpu_device *adev,
return 0;
}
-static int vcn_v2_0_dec_ring_test_ring(struct amdgpu_ring *ring)
+int vcn_v2_0_dec_ring_test_ring(struct amdgpu_ring *ring)
{
struct amdgpu_device *adev = ring->adev;
uint32_t tmp = 0;
unsigned i;
int r;
+ if (amdgpu_sriov_vf(adev))
+ return 0;
+
WREG32(adev->vcn.inst[ring->me].external.scratch9, 0xCAFEDEAD);
r = amdgpu_ring_alloc(ring, 4);
if (r)
@@ -2126,6 +1757,11 @@ static int vcn_v2_0_set_powergating_state(void *handle,
int ret;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ if (amdgpu_sriov_vf(adev)) {
+ adev->vcn.cur_state = AMD_PG_STATE_UNGATE;
+ return 0;
+ }
+
if (state == adev->vcn.cur_state)
return 0;
@@ -2139,6 +1775,211 @@ static int vcn_v2_0_set_powergating_state(void *handle,
return ret;
}
+static int vcn_v2_0_start_mmsch(struct amdgpu_device *adev,
+ struct amdgpu_mm_table *table)
+{
+ uint32_t data = 0, loop;
+ uint64_t addr = table->gpu_addr;
+ struct mmsch_v2_0_init_header *header;
+ uint32_t size;
+ int i;
+
+ header = (struct mmsch_v2_0_init_header *)table->cpu_addr;
+ size = header->header_size + header->vcn_table_size;
+
+ /* 1, write to vce_mmsch_vf_ctx_addr_lo/hi register with GPU mc addr
+ * of memory descriptor location
+ */
+ WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_ADDR_LO, lower_32_bits(addr));
+ WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_ADDR_HI, upper_32_bits(addr));
+
+ /* 2, update vmid of descriptor */
+ data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_VMID);
+ data &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK;
+ /* use domain0 for MM scheduler */
+ data |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT);
+ WREG32_SOC15(UVD, 0, mmMMSCH_VF_VMID, data);
+
+ /* 3, notify mmsch about the size of this descriptor */
+ WREG32_SOC15(UVD, 0, mmMMSCH_VF_CTX_SIZE, size);
+
+ /* 4, set resp to zero */
+ WREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP, 0);
+
+ adev->vcn.inst->ring_dec.wptr = 0;
+ adev->vcn.inst->ring_dec.wptr_old = 0;
+ vcn_v2_0_dec_ring_set_wptr(&adev->vcn.inst->ring_dec);
+
+ for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
+ adev->vcn.inst->ring_enc[i].wptr = 0;
+ adev->vcn.inst->ring_enc[i].wptr_old = 0;
+ vcn_v2_0_enc_ring_set_wptr(&adev->vcn.inst->ring_enc[i]);
+ }
+
+ /* 5, kick off the initialization and wait until
+ * VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero
+ */
+ WREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_HOST, 0x10000001);
+
+ data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP);
+ loop = 1000;
+ while ((data & 0x10000002) != 0x10000002) {
+ udelay(10);
+ data = RREG32_SOC15(UVD, 0, mmMMSCH_VF_MAILBOX_RESP);
+ loop--;
+ if (!loop)
+ break;
+ }
+
+ if (!loop) {
+ DRM_ERROR("failed to init MMSCH, " \
+ "mmMMSCH_VF_MAILBOX_RESP = 0x%08x\n", data);
+ return -EBUSY;
+ }
+
+ return 0;
+}
+
+static int vcn_v2_0_start_sriov(struct amdgpu_device *adev)
+{
+ int r;
+ uint32_t tmp;
+ struct amdgpu_ring *ring;
+ uint32_t offset, size;
+ uint32_t table_size = 0;
+ struct mmsch_v2_0_cmd_direct_write direct_wt = { {0} };
+ struct mmsch_v2_0_cmd_direct_read_modify_write direct_rd_mod_wt = { {0} };
+ struct mmsch_v2_0_cmd_end end = { {0} };
+ struct mmsch_v2_0_init_header *header;
+ uint32_t *init_table = adev->virt.mm_table.cpu_addr;
+ uint8_t i = 0;
+
+ header = (struct mmsch_v2_0_init_header *)init_table;
+ direct_wt.cmd_header.command_type = MMSCH_COMMAND__DIRECT_REG_WRITE;
+ direct_rd_mod_wt.cmd_header.command_type =
+ MMSCH_COMMAND__DIRECT_REG_READ_MODIFY_WRITE;
+ end.cmd_header.command_type = MMSCH_COMMAND__END;
+
+ if (header->vcn_table_offset == 0 && header->vcn_table_size == 0) {
+ header->version = MMSCH_VERSION;
+ header->header_size = sizeof(struct mmsch_v2_0_init_header) >> 2;
+
+ header->vcn_table_offset = header->header_size;
+
+ init_table += header->vcn_table_offset;
+
+ size = AMDGPU_GPU_PAGE_ALIGN(adev->vcn.fw->size + 4);
+
+ MMSCH_V2_0_INSERT_DIRECT_RD_MOD_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_STATUS),
+ 0xFFFFFFFF, 0x00000004);
+
+ /* mc resume*/
+ if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+ adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_lo);
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+ adev->firmware.ucode[AMDGPU_UCODE_ID_VCN].tmr_mc_addr_hi);
+ offset = 0;
+ } else {
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst->gpu_addr));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst->gpu_addr));
+ offset = size;
+ }
+
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET0),
+ 0);
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE0),
+ size);
+
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst->gpu_addr + offset));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE1_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst->gpu_addr + offset));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET1),
+ 0);
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE1),
+ AMDGPU_VCN_STACK_SIZE);
+
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_LOW),
+ lower_32_bits(adev->vcn.inst->gpu_addr + offset +
+ AMDGPU_VCN_STACK_SIZE));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_VCPU_CACHE2_64BIT_BAR_HIGH),
+ upper_32_bits(adev->vcn.inst->gpu_addr + offset +
+ AMDGPU_VCN_STACK_SIZE));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_OFFSET2),
+ 0);
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_VCPU_CACHE_SIZE2),
+ AMDGPU_VCN_CONTEXT_SIZE);
+
+ for (r = 0; r < adev->vcn.num_enc_rings; ++r) {
+ ring = &adev->vcn.inst->ring_enc[r];
+ ring->wptr = 0;
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_LO),
+ lower_32_bits(ring->gpu_addr));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_RB_BASE_HI),
+ upper_32_bits(ring->gpu_addr));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_RB_SIZE),
+ ring->ring_size / 4);
+ }
+
+ ring = &adev->vcn.inst->ring_dec;
+ ring->wptr = 0;
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
+ lower_32_bits(ring->gpu_addr));
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i,
+ mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
+ upper_32_bits(ring->gpu_addr));
+ /* force RBC into idle state */
+ tmp = order_base_2(ring->ring_size);
+ tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, tmp);
+ tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
+ tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
+ tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
+ tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
+ MMSCH_V2_0_INSERT_DIRECT_WT(
+ SOC15_REG_OFFSET(UVD, i, mmUVD_RBC_RB_CNTL), tmp);
+
+ /* add end packet */
+ tmp = sizeof(struct mmsch_v2_0_cmd_end);
+ memcpy((void *)init_table, &end, tmp);
+ table_size += (tmp / 4);
+ header->vcn_table_size = table_size;
+
+ }
+ return vcn_v2_0_start_mmsch(adev, &adev->virt.mm_table);
+}
+
static const struct amd_ip_funcs vcn_v2_0_ip_funcs = {
.name = "vcn_v2_0",
.early_init = vcn_v2_0_early_init,
@@ -2219,36 +2060,6 @@ static const struct amdgpu_ring_funcs vcn_v2_0_enc_ring_vm_funcs = {
.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
};
-static const struct amdgpu_ring_funcs vcn_v2_0_jpeg_ring_vm_funcs = {
- .type = AMDGPU_RING_TYPE_VCN_JPEG,
- .align_mask = 0xf,
- .vmhub = AMDGPU_MMHUB_0,
- .get_rptr = vcn_v2_0_jpeg_ring_get_rptr,
- .get_wptr = vcn_v2_0_jpeg_ring_get_wptr,
- .set_wptr = vcn_v2_0_jpeg_ring_set_wptr,
- .emit_frame_size =
- SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 +
- SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 +
- 8 + /* vcn_v2_0_jpeg_ring_emit_vm_flush */
- 18 + 18 + /* vcn_v2_0_jpeg_ring_emit_fence x2 vm fence */
- 8 + 16,
- .emit_ib_size = 22, /* vcn_v2_0_jpeg_ring_emit_ib */
- .emit_ib = vcn_v2_0_jpeg_ring_emit_ib,
- .emit_fence = vcn_v2_0_jpeg_ring_emit_fence,
- .emit_vm_flush = vcn_v2_0_jpeg_ring_emit_vm_flush,
- .test_ring = amdgpu_vcn_jpeg_ring_test_ring,
- .test_ib = amdgpu_vcn_jpeg_ring_test_ib,
- .insert_nop = vcn_v2_0_jpeg_ring_nop,
- .insert_start = vcn_v2_0_jpeg_ring_insert_start,
- .insert_end = vcn_v2_0_jpeg_ring_insert_end,
- .pad_ib = amdgpu_ring_generic_pad_ib,
- .begin_use = amdgpu_vcn_ring_begin_use,
- .end_use = amdgpu_vcn_ring_end_use,
- .emit_wreg = vcn_v2_0_jpeg_ring_emit_wreg,
- .emit_reg_wait = vcn_v2_0_jpeg_ring_emit_reg_wait,
- .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
-};
-
static void vcn_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev)
{
adev->vcn.inst->ring_dec.funcs = &vcn_v2_0_dec_ring_vm_funcs;
@@ -2265,12 +2076,6 @@ static void vcn_v2_0_set_enc_ring_funcs(struct amdgpu_device *adev)
DRM_INFO("VCN encode is enabled in VM mode\n");
}
-static void vcn_v2_0_set_jpeg_ring_funcs(struct amdgpu_device *adev)
-{
- adev->vcn.inst->ring_jpeg.funcs = &vcn_v2_0_jpeg_ring_vm_funcs;
- DRM_INFO("VCN jpeg decode is enabled in VM mode\n");
-}
-
static const struct amdgpu_irq_src_funcs vcn_v2_0_irq_funcs = {
.set = vcn_v2_0_set_interrupt_state,
.process = vcn_v2_0_process_interrupt,
@@ -2278,7 +2083,7 @@ static const struct amdgpu_irq_src_funcs vcn_v2_0_irq_funcs = {
static void vcn_v2_0_set_irq_funcs(struct amdgpu_device *adev)
{
- adev->vcn.inst->irq.num_types = adev->vcn.num_enc_rings + 2;
+ adev->vcn.inst->irq.num_types = adev->vcn.num_enc_rings + 1;
adev->vcn.inst->irq.funcs = &vcn_v2_0_irq_funcs;
}
generated by cgit v1.2.3-70-g09d2 (git 2.48.1) at 2025-06-01 00:56:24 +0000