From: Leo Liu Date: Fri, 3 Mar 2017 23:27:49 +0000 (-0500) Subject: drm/amdgpu: add initial vce 4.0 support for vega10 X-Git-Tag: android-x86-8.1-r1~3307^2~26^2~110 X-Git-Url: http://git.osdn.net/view?a=commitdiff_plain;h=c1dc356a116c992433c5c68e7c493b0fa3e6f9b1;p=android-x86%2Fkernel.git drm/amdgpu: add initial vce 4.0 support for vega10 Signed-off-by: Leo Liu Reviewed-by: Alex Deucher Reviewed-by: Christian König Signed-off-by: Alex Deucher --- diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index e1be2e1a10fd..cab8eecf2ecf 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -84,7 +84,8 @@ amdgpu-y += \ # add VCE block amdgpu-y += \ amdgpu_vce.o \ - vce_v3_0.o + vce_v3_0.o \ + vce_v4_0.o # add amdkfd interfaces amdgpu-y += \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index e2c06780ce49..199bc89a4d01 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -54,6 +54,8 @@ #define FIRMWARE_POLARIS11 "amdgpu/polaris11_vce.bin" #define FIRMWARE_POLARIS12 "amdgpu/polaris12_vce.bin" +#define FIRMWARE_VEGA10 "amdgpu/vega10_vce.bin" + #ifdef CONFIG_DRM_AMDGPU_CIK MODULE_FIRMWARE(FIRMWARE_BONAIRE); MODULE_FIRMWARE(FIRMWARE_KABINI); @@ -69,6 +71,8 @@ MODULE_FIRMWARE(FIRMWARE_POLARIS10); MODULE_FIRMWARE(FIRMWARE_POLARIS11); MODULE_FIRMWARE(FIRMWARE_POLARIS12); +MODULE_FIRMWARE(FIRMWARE_VEGA10); + static void amdgpu_vce_idle_work_handler(struct work_struct *work); /** @@ -123,6 +127,9 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size) case CHIP_POLARIS11: fw_name = FIRMWARE_POLARIS11; break; + case CHIP_VEGA10: + fw_name = FIRMWARE_VEGA10; + break; case CHIP_POLARIS12: fw_name = FIRMWARE_POLARIS12; break; diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c new file mode 100644 index 000000000000..74146bed5573 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c @@ -0,0 +1,894 @@ +/* + * Copyright 2016 Advanced Micro Devices, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + */ + +#include +#include +#include "amdgpu.h" +#include "amdgpu_vce.h" +#include "soc15d.h" +#include "soc15_common.h" + +#include "vega10/soc15ip.h" +#include "vega10/VCE/vce_4_0_offset.h" +#include "vega10/VCE/vce_4_0_default.h" +#include "vega10/VCE/vce_4_0_sh_mask.h" +#include "vega10/MMHUB/mmhub_1_0_offset.h" +#include "vega10/MMHUB/mmhub_1_0_sh_mask.h" + +#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK 0x02 + +#define VCE_V4_0_FW_SIZE (384 * 1024) +#define VCE_V4_0_STACK_SIZE (64 * 1024) +#define VCE_V4_0_DATA_SIZE ((16 * 1024 * AMDGPU_MAX_VCE_HANDLES) + (52 * 1024)) + +static void vce_v4_0_mc_resume(struct amdgpu_device *adev); +static void vce_v4_0_set_ring_funcs(struct amdgpu_device *adev); +static void vce_v4_0_set_irq_funcs(struct amdgpu_device *adev); + +/** + * vce_v4_0_ring_get_rptr - get read pointer + * + * @ring: amdgpu_ring pointer + * + * Returns the current hardware read pointer + */ +static uint64_t vce_v4_0_ring_get_rptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring == &adev->vce.ring[0]) + return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR)); + else if (ring == &adev->vce.ring[1]) + return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR2)); + else + return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR3)); +} + +/** + * vce_v4_0_ring_get_wptr - get write pointer + * + * @ring: amdgpu_ring pointer + * + * Returns the current hardware write pointer + */ +static uint64_t vce_v4_0_ring_get_wptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring == &adev->vce.ring[0]) + return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR)); + else if (ring == &adev->vce.ring[1]) + return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2)); + else + return RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3)); +} + +/** + * vce_v4_0_ring_set_wptr - set write pointer + * + * @ring: amdgpu_ring pointer + * + * Commits the write pointer to the hardware + */ +static void vce_v4_0_ring_set_wptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring == &adev->vce.ring[0]) + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR), + lower_32_bits(ring->wptr)); + else if (ring == &adev->vce.ring[1]) + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2), + lower_32_bits(ring->wptr)); + else + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3), + lower_32_bits(ring->wptr)); +} + +static int vce_v4_0_firmware_loaded(struct amdgpu_device *adev) +{ + int i, j; + + for (i = 0; i < 10; ++i) { + for (j = 0; j < 100; ++j) { + uint32_t status = + RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS)); + + if (status & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK) + return 0; + mdelay(10); + } + + DRM_ERROR("VCE not responding, trying to reset the ECPU!!!\n"); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET), + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(10); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET), 0, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(10); + + } + + return -ETIMEDOUT; +} + +/** + * vce_v4_0_start - start VCE block + * + * @adev: amdgpu_device pointer + * + * Setup and start the VCE block + */ +static int vce_v4_0_start(struct amdgpu_device *adev) +{ + struct amdgpu_ring *ring; + int r; + + ring = &adev->vce.ring[0]; + + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR), lower_32_bits(ring->wptr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR), lower_32_bits(ring->wptr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO), ring->gpu_addr); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI), upper_32_bits(ring->gpu_addr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE), ring->ring_size / 4); + + ring = &adev->vce.ring[1]; + + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR2), lower_32_bits(ring->wptr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR2), lower_32_bits(ring->wptr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO2), ring->gpu_addr); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI2), upper_32_bits(ring->gpu_addr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE2), ring->ring_size / 4); + + ring = &adev->vce.ring[2]; + + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_RPTR3), lower_32_bits(ring->wptr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_WPTR3), lower_32_bits(ring->wptr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_LO3), ring->gpu_addr); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_BASE_HI3), upper_32_bits(ring->gpu_addr)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_SIZE3), ring->ring_size / 4); + + vce_v4_0_mc_resume(adev); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), VCE_STATUS__JOB_BUSY_MASK, + ~VCE_STATUS__JOB_BUSY_MASK); + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CNTL), 1, ~0x200001); + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET), 0, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(100); + + r = vce_v4_0_firmware_loaded(adev); + + /* clear BUSY flag */ + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), 0, ~VCE_STATUS__JOB_BUSY_MASK); + + if (r) { + DRM_ERROR("VCE not responding, giving up!!!\n"); + return r; + } + + return 0; +} + +static int vce_v4_0_stop(struct amdgpu_device *adev) +{ + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CNTL), 0, ~0x200001); + + /* hold on ECPU */ + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SOFT_RESET), + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + + /* clear BUSY flag */ + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS), 0, ~VCE_STATUS__JOB_BUSY_MASK); + + /* Set Clock-Gating off */ + /* if (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG) + vce_v4_0_set_vce_sw_clock_gating(adev, false); + */ + + return 0; +} + +static int vce_v4_0_early_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + adev->vce.num_rings = 3; + + vce_v4_0_set_ring_funcs(adev); + vce_v4_0_set_irq_funcs(adev); + + return 0; +} + +static int vce_v4_0_sw_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_ring *ring; + unsigned size; + int r, i; + + r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VCE0, 167, &adev->vce.irq); + if (r) + return r; + + size = (VCE_V4_0_STACK_SIZE + VCE_V4_0_DATA_SIZE) * 2; + if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) + size += VCE_V4_0_FW_SIZE; + + r = amdgpu_vce_sw_init(adev, size); + 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->vce.fw->data; + adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].ucode_id = AMDGPU_UCODE_ID_VCE; + adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].fw = adev->vce.fw; + adev->firmware.fw_size += + ALIGN(le32_to_cpu(hdr->ucode_size_bytes), PAGE_SIZE); + DRM_INFO("PSP loading VCE firmware\n"); + } + + if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) { + r = amdgpu_vce_resume(adev); + if (r) + return r; + } + + for (i = 0; i < adev->vce.num_rings; i++) { + ring = &adev->vce.ring[i]; + sprintf(ring->name, "vce%d", i); + r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0); + if (r) + return r; + } + + return r; +} + +static int vce_v4_0_sw_fini(void *handle) +{ + int r; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + r = amdgpu_vce_suspend(adev); + if (r) + return r; + + r = amdgpu_vce_sw_fini(adev); + if (r) + return r; + + return r; +} + +static int vce_v4_0_hw_init(void *handle) +{ + int r, i; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + r = vce_v4_0_start(adev); + if (r) + return r; + + for (i = 0; i < adev->vce.num_rings; i++) + adev->vce.ring[i].ready = false; + + for (i = 0; i < adev->vce.num_rings; i++) { + r = amdgpu_ring_test_ring(&adev->vce.ring[i]); + if (r) + return r; + else + adev->vce.ring[i].ready = true; + } + + DRM_INFO("VCE initialized successfully.\n"); + + return 0; +} + +static int vce_v4_0_hw_fini(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + int i; + + /* vce_v4_0_wait_for_idle(handle); */ + vce_v4_0_stop(adev); + for (i = 0; i < adev->vce.num_rings; i++) + adev->vce.ring[i].ready = false; + + return 0; +} + +static int vce_v4_0_suspend(void *handle) +{ + int r; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + r = vce_v4_0_hw_fini(adev); + if (r) + return r; + + r = amdgpu_vce_suspend(adev); + if (r) + return r; + + return r; +} + +static int vce_v4_0_resume(void *handle) +{ + int r; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + r = amdgpu_vce_resume(adev); + if (r) + return r; + + r = vce_v4_0_hw_init(adev); + if (r) + return r; + + return r; +} + +static void vce_v4_0_mc_resume(struct amdgpu_device *adev) +{ + uint32_t offset, size; + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A), 0, ~(1 << 16)); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), 0x1FF000, ~0xFF9FF000); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), 0x3F, ~0x3F); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), 0x1FF); + + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL), 0x00398000); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CACHE_CTRL), 0x0, ~0x1); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL), 0); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_SWAP_CNTL1), 0); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VM_CTRL), 0); + + if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) { + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0), + (adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 8)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR0), + (adev->firmware.ucode[AMDGPU_UCODE_ID_VCE].mc_addr >> 40) & 0xff); + } else { + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR0), + (adev->vce.gpu_addr >> 8)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR0), + (adev->vce.gpu_addr >> 40) & 0xff); + } + + offset = AMDGPU_VCE_FIRMWARE_OFFSET; + size = VCE_V4_0_FW_SIZE; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET0), offset & ~0x0f000000); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE0), size); + + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR1), (adev->vce.gpu_addr >> 8)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR1), (adev->vce.gpu_addr >> 40) & 0xff); + offset = (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) ? offset + size : 0; + size = VCE_V4_0_STACK_SIZE; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET1), (offset & ~0x0f000000) | (1 << 24)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE1), size); + + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_40BIT_BAR2), (adev->vce.gpu_addr >> 8)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_VCPU_CACHE_64BIT_BAR2), (adev->vce.gpu_addr >> 40) & 0xff); + offset += size; + size = VCE_V4_0_DATA_SIZE; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_OFFSET2), (offset & ~0x0f000000) | (2 << 24)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_VCPU_CACHE_SIZE2), size); + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL2), 0x0, ~0x100); + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), + VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK, + ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK); +} + +static int vce_v4_0_set_clockgating_state(void *handle, + enum amd_clockgating_state state) +{ + /* needed for driver unload*/ + return 0; +} + +#if 0 +static bool vce_v4_0_is_idle(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 mask = 0; + + mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE0) ? 0 : SRBM_STATUS2__VCE0_BUSY_MASK; + mask |= (adev->vce.harvest_config & AMDGPU_VCE_HARVEST_VCE1) ? 0 : SRBM_STATUS2__VCE1_BUSY_MASK; + + return !(RREG32(mmSRBM_STATUS2) & mask); +} + +static int vce_v4_0_wait_for_idle(void *handle) +{ + unsigned i; + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + for (i = 0; i < adev->usec_timeout; i++) + if (vce_v4_0_is_idle(handle)) + return 0; + + return -ETIMEDOUT; +} + +#define VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK 0x00000008L /* AUTO_BUSY */ +#define VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK 0x00000010L /* RB0_BUSY */ +#define VCE_STATUS_VCPU_REPORT_RB1_BUSY_MASK 0x00000020L /* RB1_BUSY */ +#define AMDGPU_VCE_STATUS_BUSY_MASK (VCE_STATUS_VCPU_REPORT_AUTO_BUSY_MASK | \ + VCE_STATUS_VCPU_REPORT_RB0_BUSY_MASK) + +static bool vce_v4_0_check_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset = 0; + + /* According to VCE team , we should use VCE_STATUS instead + * SRBM_STATUS.VCE_BUSY bit for busy status checking. + * GRBM_GFX_INDEX.INSTANCE_INDEX is used to specify which VCE + * instance's registers are accessed + * (0 for 1st instance, 10 for 2nd instance). + * + *VCE_STATUS + *|UENC|ACPI|AUTO ACTIVE|RB1 |RB0 |RB2 | |FW_LOADED|JOB | + *|----+----+-----------+----+----+----+----------+---------+----| + *|bit8|bit7| bit6 |bit5|bit4|bit3| bit2 | bit1 |bit0| + * + * VCE team suggest use bit 3--bit 6 for busy status check + */ + mutex_lock(&adev->grbm_idx_mutex); + WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0); + if (RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) { + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1); + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1); + } + WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0x10); + if (RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS) & AMDGPU_VCE_STATUS_BUSY_MASK) { + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE0, 1); + srbm_soft_reset = REG_SET_FIELD(srbm_soft_reset, SRBM_SOFT_RESET, SOFT_RESET_VCE1, 1); + } + WREG32_FIELD(GRBM_GFX_INDEX, INSTANCE_INDEX, 0); + mutex_unlock(&adev->grbm_idx_mutex); + + if (srbm_soft_reset) { + adev->vce.srbm_soft_reset = srbm_soft_reset; + return true; + } else { + adev->vce.srbm_soft_reset = 0; + return false; + } +} + +static int vce_v4_0_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + u32 srbm_soft_reset; + + if (!adev->vce.srbm_soft_reset) + return 0; + srbm_soft_reset = adev->vce.srbm_soft_reset; + + if (srbm_soft_reset) { + u32 tmp; + + tmp = RREG32(mmSRBM_SOFT_RESET); + tmp |= srbm_soft_reset; + dev_info(adev->dev, "SRBM_SOFT_RESET=0x%08X\n", tmp); + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); + + udelay(50); + + tmp &= ~srbm_soft_reset; + WREG32(mmSRBM_SOFT_RESET, tmp); + tmp = RREG32(mmSRBM_SOFT_RESET); + + /* Wait a little for things to settle down */ + udelay(50); + } + + return 0; +} + +static int vce_v4_0_pre_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->vce.srbm_soft_reset) + return 0; + + mdelay(5); + + return vce_v4_0_suspend(adev); +} + + +static int vce_v4_0_post_soft_reset(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!adev->vce.srbm_soft_reset) + return 0; + + mdelay(5); + + return vce_v4_0_resume(adev); +} + +static void vce_v4_0_override_vce_clock_gating(struct amdgpu_device *adev, bool override) +{ + u32 tmp, data; + + tmp = data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_ARB_CTRL)); + if (override) + data |= VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK; + else + data &= ~VCE_RB_ARB_CTRL__VCE_CGTT_OVERRIDE_MASK; + + if (tmp != data) + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_RB_ARB_CTRL), data); +} + +static void vce_v4_0_set_vce_sw_clock_gating(struct amdgpu_device *adev, + bool gated) +{ + u32 data; + + /* Set Override to disable Clock Gating */ + vce_v4_0_override_vce_clock_gating(adev, true); + + /* This function enables MGCG which is controlled by firmware. + With the clocks in the gated state the core is still + accessible but the firmware will throttle the clocks on the + fly as necessary. + */ + if (gated) { + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B)); + data |= 0x1ff; + data &= ~0xef0000; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING)); + data |= 0x3ff000; + data &= ~0xffc00000; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2)); + data |= 0x2; + data &= ~0x00010000; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING)); + data |= 0x37f; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL)); + data |= VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | + 0x8; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL), data); + } else { + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B)); + data &= ~0x80010; + data |= 0xe70008; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_B), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING)); + data |= 0xffc00000; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2)); + data |= 0x10000; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING_2), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING)); + data &= ~0xffc00000; + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_REG_CLOCK_GATING), data); + + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL)); + data &= ~(VCE_UENC_DMA_DCLK_CTRL__WRDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__RDDMCLK_FORCEON_MASK | + VCE_UENC_DMA_DCLK_CTRL__REGCLK_FORCEON_MASK | + 0x8); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_DMA_DCLK_CTRL), data); + } + vce_v4_0_override_vce_clock_gating(adev, false); +} + +static void vce_v4_0_set_bypass_mode(struct amdgpu_device *adev, bool enable) +{ + u32 tmp = RREG32_SMC(ixGCK_DFS_BYPASS_CNTL); + + if (enable) + tmp |= GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK; + else + tmp &= ~GCK_DFS_BYPASS_CNTL__BYPASSECLK_MASK; + + WREG32_SMC(ixGCK_DFS_BYPASS_CNTL, tmp); +} + +static int vce_v4_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; + int i; + + if ((adev->asic_type == CHIP_POLARIS10) || + (adev->asic_type == CHIP_TONGA) || + (adev->asic_type == CHIP_FIJI)) + vce_v4_0_set_bypass_mode(adev, enable); + + if (!(adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)) + return 0; + + mutex_lock(&adev->grbm_idx_mutex); + for (i = 0; i < 2; i++) { + /* Program VCE Instance 0 or 1 if not harvested */ + if (adev->vce.harvest_config & (1 << i)) + continue; + + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, i); + + if (enable) { + /* initialize VCE_CLOCK_GATING_A: Clock ON/OFF delay */ + uint32_t data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A); + data &= ~(0xf | 0xff0); + data |= ((0x0 << 0) | (0x04 << 4)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_CLOCK_GATING_A, data); + + /* initialize VCE_UENC_CLOCK_GATING: Clock ON/OFF delay */ + data = RREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING); + data &= ~(0xf | 0xff0); + data |= ((0x0 << 0) | (0x04 << 4)); + WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_UENC_CLOCK_GATING, data); + } + + vce_v4_0_set_vce_sw_clock_gating(adev, enable); + } + + WREG32_FIELD(GRBM_GFX_INDEX, VCE_INSTANCE, 0); + mutex_unlock(&adev->grbm_idx_mutex); + + return 0; +} + +static int vce_v4_0_set_powergating_state(void *handle, + enum amd_powergating_state state) +{ + /* This doesn't actually powergate the VCE block. + * That's done in the dpm code via the SMC. This + * just re-inits the block as necessary. The actual + * gating still happens in the dpm code. We should + * revisit this when there is a cleaner line between + * the smc and the hw blocks + */ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + + if (!(adev->pg_flags & AMD_PG_SUPPORT_VCE)) + return 0; + + if (state == AMD_PG_STATE_GATE) + /* XXX do we need a vce_v4_0_stop()? */ + return 0; + else + return vce_v4_0_start(adev); +} +#endif + +static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring, + struct amdgpu_ib *ib, unsigned int vm_id, bool ctx_switch) +{ + amdgpu_ring_write(ring, VCE_CMD_IB_VM); + amdgpu_ring_write(ring, vm_id); + amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr)); + amdgpu_ring_write(ring, upper_32_bits(ib->gpu_addr)); + amdgpu_ring_write(ring, ib->length_dw); +} + +static void vce_v4_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, + u64 seq, unsigned flags) +{ + WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT); + + amdgpu_ring_write(ring, VCE_CMD_FENCE); + amdgpu_ring_write(ring, addr); + amdgpu_ring_write(ring, upper_32_bits(addr)); + amdgpu_ring_write(ring, seq); + amdgpu_ring_write(ring, VCE_CMD_TRAP); +} + +static void vce_v4_0_ring_insert_end(struct amdgpu_ring *ring) +{ + amdgpu_ring_write(ring, VCE_CMD_END); +} + +static void vce_v4_0_emit_vm_flush(struct amdgpu_ring *ring, + unsigned int vm_id, uint64_t pd_addr) +{ + unsigned eng = ring->idx; + unsigned i; + + pd_addr = pd_addr | 0x1; /* valid bit */ + /* now only use physical base address of PDE and valid */ + BUG_ON(pd_addr & 0xFFFF00000000003EULL); + + for (i = 0; i < AMDGPU_MAX_VMHUBS; ++i) { + struct amdgpu_vmhub *hub = &ring->adev->vmhub[i]; + uint32_t req = hub->get_invalidate_req(vm_id); + + amdgpu_ring_write(ring, VCE_CMD_REG_WRITE); + amdgpu_ring_write(ring, + (hub->ctx0_ptb_addr_hi32 + vm_id * 2) << 2); + amdgpu_ring_write(ring, upper_32_bits(pd_addr)); + + amdgpu_ring_write(ring, VCE_CMD_REG_WRITE); + amdgpu_ring_write(ring, + (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2); + amdgpu_ring_write(ring, lower_32_bits(pd_addr)); + + amdgpu_ring_write(ring, VCE_CMD_REG_WAIT); + amdgpu_ring_write(ring, + (hub->ctx0_ptb_addr_lo32 + vm_id * 2) << 2); + amdgpu_ring_write(ring, 0xffffffff); + amdgpu_ring_write(ring, lower_32_bits(pd_addr)); + + /* flush TLB */ + amdgpu_ring_write(ring, VCE_CMD_REG_WRITE); + amdgpu_ring_write(ring, (hub->vm_inv_eng0_req + eng) << 2); + amdgpu_ring_write(ring, req); + + /* wait for flush */ + amdgpu_ring_write(ring, VCE_CMD_REG_WAIT); + amdgpu_ring_write(ring, (hub->vm_inv_eng0_ack + eng) << 2); + amdgpu_ring_write(ring, 1 << vm_id); + amdgpu_ring_write(ring, 1 << vm_id); + } +} + +static int vce_v4_0_set_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned type, + enum amdgpu_interrupt_state state) +{ + uint32_t val = 0; + + if (state == AMDGPU_IRQ_STATE_ENABLE) + val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK; + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), val, + ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK); + return 0; +} + +static int vce_v4_0_process_interrupt(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + DRM_DEBUG("IH: VCE\n"); + + WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_STATUS), + VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK, + ~VCE_SYS_INT_STATUS__VCE_SYS_INT_TRAP_INTERRUPT_INT_MASK); + + switch (entry->src_data[0]) { + case 0: + case 1: + case 2: + amdgpu_fence_process(&adev->vce.ring[entry->src_data[0]]); + break; + default: + DRM_ERROR("Unhandled interrupt: %d %d\n", + entry->src_id, entry->src_data[0]); + break; + } + + return 0; +} + +const struct amd_ip_funcs vce_v4_0_ip_funcs = { + .name = "vce_v4_0", + .early_init = vce_v4_0_early_init, + .late_init = NULL, + .sw_init = vce_v4_0_sw_init, + .sw_fini = vce_v4_0_sw_fini, + .hw_init = vce_v4_0_hw_init, + .hw_fini = vce_v4_0_hw_fini, + .suspend = vce_v4_0_suspend, + .resume = vce_v4_0_resume, + .is_idle = NULL /* vce_v4_0_is_idle */, + .wait_for_idle = NULL /* vce_v4_0_wait_for_idle */, + .check_soft_reset = NULL /* vce_v4_0_check_soft_reset */, + .pre_soft_reset = NULL /* vce_v4_0_pre_soft_reset */, + .soft_reset = NULL /* vce_v4_0_soft_reset */, + .post_soft_reset = NULL /* vce_v4_0_post_soft_reset */, + .set_clockgating_state = vce_v4_0_set_clockgating_state, + .set_powergating_state = NULL /* vce_v4_0_set_powergating_state */, +}; + +static const struct amdgpu_ring_funcs vce_v4_0_ring_vm_funcs = { + .type = AMDGPU_RING_TYPE_VCE, + .align_mask = 0x3f, + .nop = VCE_CMD_NO_OP, + .support_64bit_ptrs = false, + .get_rptr = vce_v4_0_ring_get_rptr, + .get_wptr = vce_v4_0_ring_get_wptr, + .set_wptr = vce_v4_0_ring_set_wptr, + .parse_cs = amdgpu_vce_ring_parse_cs_vm, + .emit_frame_size = + 17 * AMDGPU_MAX_VMHUBS + /* vce_v4_0_emit_vm_flush */ + 5 + 5 + /* amdgpu_vce_ring_emit_fence x2 vm fence */ + 1, /* vce_v4_0_ring_insert_end */ + .emit_ib_size = 5, /* vce_v4_0_ring_emit_ib */ + .emit_ib = vce_v4_0_ring_emit_ib, + .emit_vm_flush = vce_v4_0_emit_vm_flush, + .emit_fence = vce_v4_0_ring_emit_fence, + .test_ring = amdgpu_vce_ring_test_ring, + .test_ib = amdgpu_vce_ring_test_ib, + .insert_nop = amdgpu_ring_insert_nop, + .insert_end = vce_v4_0_ring_insert_end, + .pad_ib = amdgpu_ring_generic_pad_ib, + .begin_use = amdgpu_vce_ring_begin_use, + .end_use = amdgpu_vce_ring_end_use, +}; + +static void vce_v4_0_set_ring_funcs(struct amdgpu_device *adev) +{ + int i; + + for (i = 0; i < adev->vce.num_rings; i++) + adev->vce.ring[i].funcs = &vce_v4_0_ring_vm_funcs; + DRM_INFO("VCE enabled in VM mode\n"); +} + +static const struct amdgpu_irq_src_funcs vce_v4_0_irq_funcs = { + .set = vce_v4_0_set_interrupt_state, + .process = vce_v4_0_process_interrupt, +}; + +static void vce_v4_0_set_irq_funcs(struct amdgpu_device *adev) +{ + adev->vce.irq.num_types = 1; + adev->vce.irq.funcs = &vce_v4_0_irq_funcs; +}; + +const struct amdgpu_ip_block_version vce_v4_0_ip_block = +{ + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 4, + .minor = 0, + .rev = 0, + .funcs = &vce_v4_0_ip_funcs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.h b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.h new file mode 100644 index 000000000000..a32beda6a473 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.h @@ -0,0 +1,29 @@ +/* + * Copyright 2016 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __VCE_V4_0_H__ +#define __VCE_V4_0_H__ + +extern const struct amdgpu_ip_block_version vce_v4_0_ip_block; + +#endif