microsoft
/
WSL2-Linux-Kernel
зеркало из https://github.com/microsoft/WSL2-Linux-Kernel.git
1
0
Форкнуть 0
Код Задачи Пакеты Проекты Релизы Вики Активность

Merge tag 'drm-next-5.3-2019-06-25' of git://people.freedesktop.org/~agd5f/linux into drm-next 

drm-next-5.3-2019-06-25:

Merge drm-next

amdgpu:
- SR-IOV L1 policy fixes
- Removed no longer needed vram_page_split module parameter
- Add module parameter to override default ABM level
- Gamma fixes
- No need to check return values for debugfs
- Improve HMM error handling
- Avoid possible OOM situations when lots of thread are submitting with
  memory contention
- Improve hw i2c access abritration
- DSC (Display Stream Compression) support in DC
- Initial navi10 support
  * DC support
  * GFX/Compute support
  * SDMA support
  * Power Management support
  * VCN support
- Static checker fixes
- Misc cleanups
- fix long udelay on arm

amdkfd:
- Implement priority controls for gfx9
- Enable VEGAM
- Rework mqd allocation and init
- Circular locking fix
- Fix SDMA queue allocation race condition
- No need to check return values for debugfs
- Add proc style process information
- Initial navi10 support

radeon:
- No need to check return values for debugfs

UAPI changes:
- GDDR6 added to vram type query
- New Navi10 details added gpu info query
- Navi family added to asic family query

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190625195520.3817-1-alexander.deucher@amd.com
This commit is contained in:
Dave Airlie 2019-06-27 12:33:56 +10:00
Родитель b22342ea07 a201b6ac20
Коммит 14808a12bd
358 изменённых файлов: 414200 добавлений и 2946 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

30
drivers/gpu/drm/amd/amdgpu/Makefile
Просмотреть файл

@ -54,7 +54,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \
amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_atomfirmware.o \
amdgpu_vf_error.o amdgpu_sched.o amdgpu_debugfs.o amdgpu_ids.o \
amdgpu_gmc.o amdgpu_xgmi.o amdgpu_csa.o amdgpu_ras.o amdgpu_vm_cpu.o \
amdgpu_vm_sdma.o
amdgpu_vm_sdma.o amdgpu_pmu.o amdgpu_discovery.o
# add asic specific block
amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
@ -64,7 +64,7 @@ amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce
amdgpu-y += \
vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \
vega20_reg_init.o nbio_v7_4.o
vega20_reg_init.o nbio_v7_4.o nbio_v2_3.o nv.o navi10_reg_init.o
# add DF block
amdgpu-y += \
@ -75,7 +75,8 @@ amdgpu-y += \
amdgpu-y += \
gmc_v7_0.o \
gmc_v8_0.o \
gfxhub_v1_0.o mmhub_v1_0.o gmc_v9_0.o gfxhub_v1_1.o
gfxhub_v1_0.o mmhub_v1_0.o gmc_v9_0.o gfxhub_v1_1.o \
gfxhub_v2_0.o mmhub_v2_0.o gmc_v10_0.o
# add IH block
amdgpu-y += \
@ -84,7 +85,8 @@ amdgpu-y += \
iceland_ih.o \
tonga_ih.o \
cz_ih.o \
vega10_ih.o
vega10_ih.o \
navi10_ih.o
# add PSP block
amdgpu-y += \
@ -108,14 +110,20 @@ amdgpu-y += \
amdgpu_gfx.o \
amdgpu_rlc.o \
gfx_v8_0.o \
gfx_v9_0.o
gfx_v9_0.o \
gfx_v10_0.o
# add async DMA block
amdgpu-y += \
amdgpu_sdma.o \
sdma_v2_4.o \
sdma_v3_0.o \
sdma_v4_0.o
sdma_v4_0.o \
sdma_v5_0.o
# add MES block
amdgpu-y += \
mes_v10_1.o
# add UVD block
amdgpu-y += \
@ -133,7 +141,12 @@ amdgpu-y += \
# add VCN block
amdgpu-y += \
amdgpu_vcn.o \
vcn_v1_0.o
vcn_v1_0.o \
vcn_v2_0.o
# add ATHUB block
amdgpu-y += \
athub_v2_0.o
# add amdkfd interfaces
amdgpu-y += amdgpu_amdkfd.o
@ -146,7 +159,8 @@ amdgpu-y += \
amdgpu_amdkfd_fence.o \
amdgpu_amdkfd_gpuvm.o \
amdgpu_amdkfd_gfx_v8.o \
amdgpu_amdkfd_gfx_v9.o
amdgpu_amdkfd_gfx_v9.o \
amdgpu_amdkfd_gfx_v10.o
ifneq ($(CONFIG_DRM_AMDGPU_CIK),)
amdgpu-y += amdgpu_amdkfd_gfx_v7.o

40
drivers/gpu/drm/amd/amdgpu/amdgpu.h
Просмотреть файл

@ -84,6 +84,8 @@
#include "amdgpu_doorbell.h"
#include "amdgpu_amdkfd.h"
#include "amdgpu_smu.h"
#include "amdgpu_discovery.h"
#include "amdgpu_mes.h"
#define MAX_GPU_INSTANCE 16
@ -142,7 +144,6 @@ extern uint amdgpu_sdma_phase_quantum;
extern char *amdgpu_disable_cu;
extern char *amdgpu_virtual_display;
extern uint amdgpu_pp_feature_mask;
extern int amdgpu_vram_page_split;
extern int amdgpu_ngg;
extern int amdgpu_prim_buf_per_se;
extern int amdgpu_pos_buf_per_se;
@ -155,9 +156,14 @@ extern int amdgpu_gpu_recovery;
extern int amdgpu_emu_mode;
extern uint amdgpu_smu_memory_pool_size;
extern uint amdgpu_dc_feature_mask;
extern uint amdgpu_dm_abm_level;
extern struct amdgpu_mgpu_info mgpu_info;
extern int amdgpu_ras_enable;
extern uint amdgpu_ras_mask;
extern int amdgpu_async_gfx_ring;
extern int amdgpu_mcbp;
extern int amdgpu_discovery;
extern int amdgpu_mes;
#ifdef CONFIG_DRM_AMDGPU_SI
extern int amdgpu_si_support;
@ -213,7 +219,8 @@ struct amdgpu_atif;
struct kfd_vm_fault_info;
enum amdgpu_cp_irq {
AMDGPU_CP_IRQ_GFX_EOP = 0,
AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP = 0,
AMDGPU_CP_IRQ_GFX_ME0_PIPE1_EOP,
AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP,
AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE1_EOP,
AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE2_EOP,
@ -659,6 +666,8 @@ struct amdgpu_nbio_funcs {
u32 (*get_memsize)(struct amdgpu_device *adev);
void (*sdma_doorbell_range)(struct amdgpu_device *adev, int instance,
bool use_doorbell, int doorbell_index, int doorbell_size);
void (*vcn_doorbell_range)(struct amdgpu_device *adev, bool use_doorbell,
int doorbell_index);
void (*enable_doorbell_aperture)(struct amdgpu_device *adev,
bool enable);
void (*enable_doorbell_selfring_aperture)(struct amdgpu_device *adev,
@ -678,7 +687,7 @@ struct amdgpu_nbio_funcs {
};
struct amdgpu_df_funcs {
void (*init)(struct amdgpu_device *adev);
void (*sw_init)(struct amdgpu_device *adev);
void (*enable_broadcast_mode)(struct amdgpu_device *adev,
bool enable);
u32 (*get_fb_channel_number)(struct amdgpu_device *adev);
@ -729,6 +738,7 @@ struct amd_powerplay {
};
#define AMDGPU_RESET_MAGIC_NUM 64
#define AMDGPU_MAX_DF_PERFMONS 4
struct amdgpu_device {
struct device *dev;
struct drm_device *ddev;
@ -755,6 +765,7 @@ struct amdgpu_device {
struct amdgpu_debugfs debugfs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
unsigned debugfs_count;
#if defined(CONFIG_DEBUG_FS)
struct dentry *debugfs_preempt;
struct dentry *debugfs_regs[AMDGPU_DEBUGFS_MAX_COMPONENTS];
#endif
struct amdgpu_atif *atif;
@ -905,6 +916,13 @@ struct amdgpu_device {
/* display related functionality */
struct amdgpu_display_manager dm;
/* discovery */
uint8_t *discovery;
/* mes */
bool enable_mes;
struct amdgpu_mes mes;
struct amdgpu_ip_block ip_blocks[AMDGPU_MAX_IP_NUM];
int num_ip_blocks;
struct mutex mn_lock;
@ -959,6 +977,7 @@ struct amdgpu_device {
long compute_timeout;
uint64_t unique_id;
uint64_t df_perfmon_config_assign_mask[AMDGPU_MAX_DF_PERFMONS];
};
static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)
@ -1198,4 +1217,19 @@ static inline int amdgpu_dm_display_resume(struct amdgpu_device *adev) { return
#endif
#include "amdgpu_object.h"
/* used by df_v3_6.c and amdgpu_pmu.c */
#define AMDGPU_PMU_ATTR(_name, _object) \
static ssize_t \
_name##_show(struct device *dev, \
struct device_attribute *attr, \
char *page) \
{ \
BUILD_BUG_ON(sizeof(_object) >= PAGE_SIZE - 1); \
return sprintf(page, _object "\n"); \
} \
\
static struct device_attribute pmu_attr_##_name = __ATTR_RO(_name)
#endif

20
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
Просмотреть файл

@ -78,6 +78,7 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
case CHIP_POLARIS10:
case CHIP_POLARIS11:
case CHIP_POLARIS12:
case CHIP_VEGAM:
kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
break;
case CHIP_VEGA10:
@ -86,6 +87,9 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
case CHIP_RAVEN:
kfd2kgd = amdgpu_amdkfd_gfx_9_0_get_functions();
break;
case CHIP_NAVI10:
kfd2kgd = amdgpu_amdkfd_gfx_10_0_get_functions();
break;
default:
dev_info(adev->dev, "kfd not supported on this ASIC\n");
return;
@ -158,7 +162,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
/* remove the KIQ bit as well */
if (adev->gfx.kiq.ring.sched.ready)
clear_bit(amdgpu_gfx_queue_to_bit(adev,
clear_bit(amdgpu_gfx_mec_queue_to_bit(adev,
adev->gfx.kiq.ring.me - 1,
adev->gfx.kiq.ring.pipe,
adev->gfx.kiq.ring.queue),
@ -436,9 +440,12 @@ void amdgpu_amdkfd_get_local_mem_info(struct kgd_dev *kgd,
if (amdgpu_sriov_vf(adev))
mem_info->mem_clk_max = adev->clock.default_mclk / 100;
else if (adev->powerplay.pp_funcs)
mem_info->mem_clk_max = amdgpu_dpm_get_mclk(adev, false) / 100;
else
else if (adev->powerplay.pp_funcs) {
if (amdgpu_emu_mode == 1)
mem_info->mem_clk_max = 0;
else
mem_info->mem_clk_max = amdgpu_dpm_get_mclk(adev, false) / 100;
} else
mem_info->mem_clk_max = 100;
}
@ -701,6 +708,11 @@ struct kfd2kgd_calls *amdgpu_amdkfd_gfx_9_0_get_functions(void)
return NULL;
}
struct kfd2kgd_calls *amdgpu_amdkfd_gfx_10_0_get_functions(void)
{
return NULL;
}
struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, struct pci_dev *pdev,
const struct kfd2kgd_calls *f2g)
{

1
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
Просмотреть файл

@ -139,6 +139,7 @@ void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle);
struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void);
struct kfd2kgd_calls *amdgpu_amdkfd_gfx_8_0_get_functions(void);
struct kfd2kgd_calls *amdgpu_amdkfd_gfx_9_0_get_functions(void);
struct kfd2kgd_calls *amdgpu_amdkfd_gfx_10_0_get_functions(void);
bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid);

975
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c Normal file
Просмотреть файл

@ -0,0 +1,975 @@
/*
* Copyright 2019 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.
*/
#undef pr_fmt
#define pr_fmt(fmt) "kfd2kgd: " fmt
#include <linux/module.h>
#include <linux/fdtable.h>
#include <linux/uaccess.h>
#include <linux/firmware.h>
#include <linux/mmu_context.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_amdkfd.h"
#include "amdgpu_ucode.h"
#include "soc15_hw_ip.h"
#include "gc/gc_10_1_0_offset.h"
#include "gc/gc_10_1_0_sh_mask.h"
#include "navi10_enum.h"
#include "athub/athub_2_0_0_offset.h"
#include "athub/athub_2_0_0_sh_mask.h"
#include "oss/osssys_5_0_0_offset.h"
#include "oss/osssys_5_0_0_sh_mask.h"
#include "soc15_common.h"
#include "v10_structs.h"
#include "nv.h"
#include "nvd.h"
enum hqd_dequeue_request_type {
NO_ACTION = 0,
DRAIN_PIPE,
RESET_WAVES,
SAVE_WAVES
};
/*
* Register access functions
*/
static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
uint32_t sh_mem_config,
uint32_t sh_mem_ape1_base, uint32_t sh_mem_ape1_limit,
uint32_t sh_mem_bases);
static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
unsigned int vmid);
static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id);
static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
uint32_t queue_id, uint32_t __user *wptr,
uint32_t wptr_shift, uint32_t wptr_mask,
struct mm_struct *mm);
static int kgd_hqd_dump(struct kgd_dev *kgd,
uint32_t pipe_id, uint32_t queue_id,
uint32_t (**dump)[2], uint32_t *n_regs);
static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
uint32_t __user *wptr, struct mm_struct *mm);
static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
uint32_t engine_id, uint32_t queue_id,
uint32_t (**dump)[2], uint32_t *n_regs);
static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
uint32_t pipe_id, uint32_t queue_id);
static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd);
static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
enum kfd_preempt_type reset_type,
unsigned int utimeout, uint32_t pipe_id,
uint32_t queue_id);
static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
unsigned int utimeout);
#if 0
static uint32_t get_watch_base_addr(struct amdgpu_device *adev);
#endif
static int kgd_address_watch_disable(struct kgd_dev *kgd);
static int kgd_address_watch_execute(struct kgd_dev *kgd,
unsigned int watch_point_id,
uint32_t cntl_val,
uint32_t addr_hi,
uint32_t addr_lo);
static int kgd_wave_control_execute(struct kgd_dev *kgd,
uint32_t gfx_index_val,
uint32_t sq_cmd);
static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
unsigned int watch_point_id,
unsigned int reg_offset);
static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
uint8_t vmid);
static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
uint8_t vmid);
static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
uint64_t page_table_base);
static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid);
static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid);
/* Because of REG_GET_FIELD() being used, we put this function in the
* asic specific file.
*/
static int amdgpu_amdkfd_get_tile_config(struct kgd_dev *kgd,
struct tile_config *config)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
config->gb_addr_config = adev->gfx.config.gb_addr_config;
#if 0
/* TODO - confirm REG_GET_FIELD x2, should be OK as is... but
* MC_ARB_RAMCFG register doesn't exist on Vega10 - initial amdgpu
* changes commented out related code, doing the same here for now but
* need to sync with Ken et al
*/
config->num_banks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
MC_ARB_RAMCFG, NOOFBANK);
config->num_ranks = REG_GET_FIELD(adev->gfx.config.mc_arb_ramcfg,
MC_ARB_RAMCFG, NOOFRANKS);
#endif
config->tile_config_ptr = adev->gfx.config.tile_mode_array;
config->num_tile_configs =
ARRAY_SIZE(adev->gfx.config.tile_mode_array);
config->macro_tile_config_ptr =
adev->gfx.config.macrotile_mode_array;
config->num_macro_tile_configs =
ARRAY_SIZE(adev->gfx.config.macrotile_mode_array);
return 0;
}
static const struct kfd2kgd_calls kfd2kgd = {
.program_sh_mem_settings = kgd_program_sh_mem_settings,
.set_pasid_vmid_mapping = kgd_set_pasid_vmid_mapping,
.init_interrupts = kgd_init_interrupts,
.hqd_load = kgd_hqd_load,
.hqd_sdma_load = kgd_hqd_sdma_load,
.hqd_dump = kgd_hqd_dump,
.hqd_sdma_dump = kgd_hqd_sdma_dump,
.hqd_is_occupied = kgd_hqd_is_occupied,
.hqd_sdma_is_occupied = kgd_hqd_sdma_is_occupied,
.hqd_destroy = kgd_hqd_destroy,
.hqd_sdma_destroy = kgd_hqd_sdma_destroy,
.address_watch_disable = kgd_address_watch_disable,
.address_watch_execute = kgd_address_watch_execute,
.wave_control_execute = kgd_wave_control_execute,
.address_watch_get_offset = kgd_address_watch_get_offset,
.get_atc_vmid_pasid_mapping_pasid =
get_atc_vmid_pasid_mapping_pasid,
.get_atc_vmid_pasid_mapping_valid =
get_atc_vmid_pasid_mapping_valid,
.invalidate_tlbs = invalidate_tlbs,
.invalidate_tlbs_vmid = invalidate_tlbs_vmid,
.set_vm_context_page_table_base = set_vm_context_page_table_base,
.get_tile_config = amdgpu_amdkfd_get_tile_config,
};
struct kfd2kgd_calls *amdgpu_amdkfd_gfx_10_0_get_functions()
{
return (struct kfd2kgd_calls *)&kfd2kgd;
}
static inline struct amdgpu_device *get_amdgpu_device(struct kgd_dev *kgd)
{
return (struct amdgpu_device *)kgd;
}
static void lock_srbm(struct kgd_dev *kgd, uint32_t mec, uint32_t pipe,
uint32_t queue, uint32_t vmid)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
mutex_lock(&adev->srbm_mutex);
nv_grbm_select(adev, mec, pipe, queue, vmid);
}
static void unlock_srbm(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
nv_grbm_select(adev, 0, 0, 0, 0);
mutex_unlock(&adev->srbm_mutex);
}
static void acquire_queue(struct kgd_dev *kgd, uint32_t pipe_id,
uint32_t queue_id)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint32_t mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
uint32_t pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
lock_srbm(kgd, mec, pipe, queue_id, 0);
}
static uint32_t get_queue_mask(struct amdgpu_device *adev,
uint32_t pipe_id, uint32_t queue_id)
{
unsigned int bit = (pipe_id * adev->gfx.mec.num_queue_per_pipe +
queue_id) & 31;
return ((uint32_t)1) << bit;
}
static void release_queue(struct kgd_dev *kgd)
{
unlock_srbm(kgd);
}
static void kgd_program_sh_mem_settings(struct kgd_dev *kgd, uint32_t vmid,
uint32_t sh_mem_config,
uint32_t sh_mem_ape1_base,
uint32_t sh_mem_ape1_limit,
uint32_t sh_mem_bases)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
lock_srbm(kgd, 0, 0, 0, vmid);
WREG32(SOC15_REG_OFFSET(GC, 0, mmSH_MEM_CONFIG), sh_mem_config);
WREG32(SOC15_REG_OFFSET(GC, 0, mmSH_MEM_BASES), sh_mem_bases);
/* APE1 no longer exists on GFX9 */
unlock_srbm(kgd);
}
static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
unsigned int vmid)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
/*
* We have to assume that there is no outstanding mapping.
* The ATC_VMID_PASID_MAPPING_UPDATE_STATUS bit could be 0 because
* a mapping is in progress or because a mapping finished
* and the SW cleared it.
* So the protocol is to always wait & clear.
*/
uint32_t pasid_mapping = (pasid == 0) ? 0 : (uint32_t)pasid |
ATC_VMID0_PASID_MAPPING__VALID_MASK;
pr_debug("pasid 0x%x vmid %d, reg value %x\n", pasid, vmid, pasid_mapping);
/*
* need to do this twice, once for gfx and once for mmhub
* for ATC add 16 to VMID for mmhub, for IH different registers.
* ATC_VMID0..15 registers are separate from ATC_VMID16..31.
*/
pr_debug("ATHUB, reg %x\n", SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + vmid);
WREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING) + vmid,
pasid_mapping);
#if 0
/* TODO: uncomment this code when the hardware support is ready. */
while (!(RREG32(SOC15_REG_OFFSET(
ATHUB, 0,
mmATC_VMID_PASID_MAPPING_UPDATE_STATUS)) &
(1U << vmid)))
cpu_relax();
pr_debug("ATHUB mapping update finished\n");
WREG32(SOC15_REG_OFFSET(ATHUB, 0,
mmATC_VMID_PASID_MAPPING_UPDATE_STATUS),
1U << vmid);
#endif
/* Mapping vmid to pasid also for IH block */
pr_debug("update mapping for IH block and mmhub");
WREG32(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid,
pasid_mapping);
return 0;
}
/* TODO - RING0 form of field is obsolete, seems to date back to SI
* but still works
*/
static int kgd_init_interrupts(struct kgd_dev *kgd, uint32_t pipe_id)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint32_t mec;
uint32_t pipe;
mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
lock_srbm(kgd, mec, pipe, 0, 0);
WREG32(SOC15_REG_OFFSET(GC, 0, mmCPC_INT_CNTL),
CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
unlock_srbm(kgd);
return 0;
}
static uint32_t get_sdma_base_addr(struct amdgpu_device *adev,
unsigned int engine_id,
unsigned int queue_id)
{
uint32_t base[2] = {
SOC15_REG_OFFSET(SDMA0, 0,
mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL,
/* On gfx10, mmSDMA1_xxx registers are defined NOT based
* on SDMA1 base address (dw 0x1860) but based on SDMA0
* base address (dw 0x1260). Therefore use mmSDMA0_RLC0_RB_CNTL
* instead of mmSDMA1_RLC0_RB_CNTL for the base address calc
* below
*/
SOC15_REG_OFFSET(SDMA1, 0,
mmSDMA1_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL
};
uint32_t retval;
retval = base[engine_id] + queue_id * (mmSDMA0_RLC1_RB_CNTL -
mmSDMA0_RLC0_RB_CNTL);
pr_debug("sdma base address: 0x%x\n", retval);
return retval;
}
#if 0
static uint32_t get_watch_base_addr(struct amdgpu_device *adev)
{
uint32_t retval = SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) -
mmTCP_WATCH0_ADDR_H;
pr_debug("kfd: reg watch base address: 0x%x\n", retval);
return retval;
}
#endif
static inline struct v10_compute_mqd *get_mqd(void *mqd)
{
return (struct v10_compute_mqd *)mqd;
}
static inline struct v10_sdma_mqd *get_sdma_mqd(void *mqd)
{
return (struct v10_sdma_mqd *)mqd;
}
static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
uint32_t queue_id, uint32_t __user *wptr,
uint32_t wptr_shift, uint32_t wptr_mask,
struct mm_struct *mm)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
struct v10_compute_mqd *m;
uint32_t *mqd_hqd;
uint32_t reg, hqd_base, data;
m = get_mqd(mqd);
pr_debug("Load hqd of pipe %d queue %d\n", pipe_id, queue_id);
acquire_queue(kgd, pipe_id, queue_id);
/* HIQ is set during driver init period with vmid set to 0*/
if (m->cp_hqd_vmid == 0) {
uint32_t value, mec, pipe;
mec = (pipe_id / adev->gfx.mec.num_pipe_per_mec) + 1;
pipe = (pipe_id % adev->gfx.mec.num_pipe_per_mec);
pr_debug("kfd: set HIQ, mec:%d, pipe:%d, queue:%d.\n",
mec, pipe, queue_id);
value = RREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CP_SCHEDULERS));
value = REG_SET_FIELD(value, RLC_CP_SCHEDULERS, scheduler1,
((mec << 5) | (pipe << 3) | queue_id | 0x80));
WREG32(SOC15_REG_OFFSET(GC, 0, mmRLC_CP_SCHEDULERS), value);
}
/* HQD registers extend from CP_MQD_BASE_ADDR to CP_HQD_EOP_WPTR_MEM. */
mqd_hqd = &m->cp_mqd_base_addr_lo;
hqd_base = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
for (reg = hqd_base;
reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
WREG32(reg, mqd_hqd[reg - hqd_base]);
/* Activate doorbell logic before triggering WPTR poll. */
data = REG_SET_FIELD(m->cp_hqd_pq_doorbell_control,
CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_EN, 1);
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_DOORBELL_CONTROL), data);
if (wptr) {
/* Don't read wptr with get_user because the user
* context may not be accessible (if this function
* runs in a work queue). Instead trigger a one-shot
* polling read from memory in the CP. This assumes
* that wptr is GPU-accessible in the queue's VMID via
* ATC or SVM. WPTR==RPTR before starting the poll so
* the CP starts fetching new commands from the right
* place.
*
* Guessing a 64-bit WPTR from a 32-bit RPTR is a bit
* tricky. Assume that the queue didn't overflow. The
* number of valid bits in the 32-bit RPTR depends on
* the queue size. The remaining bits are taken from
* the saved 64-bit WPTR. If the WPTR wrapped, add the
* queue size.
*/
uint32_t queue_size =
2 << REG_GET_FIELD(m->cp_hqd_pq_control,
CP_HQD_PQ_CONTROL, QUEUE_SIZE);
uint64_t guessed_wptr = m->cp_hqd_pq_rptr & (queue_size - 1);
if ((m->cp_hqd_pq_wptr_lo & (queue_size - 1)) < guessed_wptr)
guessed_wptr += queue_size;
guessed_wptr += m->cp_hqd_pq_wptr_lo & ~(queue_size - 1);
guessed_wptr += (uint64_t)m->cp_hqd_pq_wptr_hi << 32;
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_LO),
lower_32_bits(guessed_wptr));
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI),
upper_32_bits(guessed_wptr));
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR),
lower_32_bits((uint64_t)wptr));
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_POLL_ADDR_HI),
upper_32_bits((uint64_t)wptr));
pr_debug("%s setting CP_PQ_WPTR_POLL_CNTL1 to %x\n", __func__, get_queue_mask(adev, pipe_id, queue_id));
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_PQ_WPTR_POLL_CNTL1),
get_queue_mask(adev, pipe_id, queue_id));
}
/* Start the EOP fetcher */
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_EOP_RPTR),
REG_SET_FIELD(m->cp_hqd_eop_rptr,
CP_HQD_EOP_RPTR, INIT_FETCHER, 1));
data = REG_SET_FIELD(m->cp_hqd_active, CP_HQD_ACTIVE, ACTIVE, 1);
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_ACTIVE), data);
release_queue(kgd);
return 0;
}
static int kgd_hqd_dump(struct kgd_dev *kgd,
uint32_t pipe_id, uint32_t queue_id,
uint32_t (**dump)[2], uint32_t *n_regs)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint32_t i = 0, reg;
#define HQD_N_REGS 56
#define DUMP_REG(addr) do { \
if (WARN_ON_ONCE(i >= HQD_N_REGS)) \
break; \
(*dump)[i][0] = (addr) << 2; \
(*dump)[i++][1] = RREG32(addr); \
} while (0)
*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
if (*dump == NULL)
return -ENOMEM;
acquire_queue(kgd, pipe_id, queue_id);
for (reg = SOC15_REG_OFFSET(GC, 0, mmCP_MQD_BASE_ADDR);
reg <= SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_WPTR_HI); reg++)
DUMP_REG(reg);
release_queue(kgd);
WARN_ON_ONCE(i != HQD_N_REGS);
*n_regs = i;
return 0;
}
static int kgd_hqd_sdma_load(struct kgd_dev *kgd, void *mqd,
uint32_t __user *wptr, struct mm_struct *mm)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
struct v10_sdma_mqd *m;
uint32_t sdma_base_addr, sdmax_gfx_context_cntl;
unsigned long end_jiffies;
uint32_t data;
uint64_t data64;
uint64_t __user *wptr64 = (uint64_t __user *)wptr;
m = get_sdma_mqd(mqd);
sdma_base_addr = get_sdma_base_addr(adev, m->sdma_engine_id,
m->sdma_queue_id);
pr_debug("sdma load base addr %x for engine %d, queue %d\n", sdma_base_addr, m->sdma_engine_id, m->sdma_queue_id);
sdmax_gfx_context_cntl = m->sdma_engine_id ?
SOC15_REG_OFFSET(SDMA1, 0, mmSDMA1_GFX_CONTEXT_CNTL) :
SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_GFX_CONTEXT_CNTL);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
m->sdmax_rlcx_rb_cntl & (~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK));
end_jiffies = msecs_to_jiffies(2000) + jiffies;
while (true) {
data = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
if (data & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
break;
if (time_after(jiffies, end_jiffies))
return -ETIME;
usleep_range(500, 1000);
}
data = RREG32(sdmax_gfx_context_cntl);
data = REG_SET_FIELD(data, SDMA0_GFX_CONTEXT_CNTL,
RESUME_CTX, 0);
WREG32(sdmax_gfx_context_cntl, data);
WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL_OFFSET,
m->sdmax_rlcx_doorbell_offset);
data = REG_SET_FIELD(m->sdmax_rlcx_doorbell, SDMA0_RLC0_DOORBELL,
ENABLE, 1);
WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, data);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR, m->sdmax_rlcx_rb_rptr);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_HI,
m->sdmax_rlcx_rb_rptr_hi);
WREG32(sdma_base_addr + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 1);
if (read_user_wptr(mm, wptr64, data64)) {
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR,
lower_32_bits(data64));
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR_HI,
upper_32_bits(data64));
} else {
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR,
m->sdmax_rlcx_rb_rptr);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_WPTR_HI,
m->sdmax_rlcx_rb_rptr_hi);
}
WREG32(sdma_base_addr + mmSDMA0_RLC0_MINOR_PTR_UPDATE, 0);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE, m->sdmax_rlcx_rb_base);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_BASE_HI,
m->sdmax_rlcx_rb_base_hi);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_LO,
m->sdmax_rlcx_rb_rptr_addr_lo);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_ADDR_HI,
m->sdmax_rlcx_rb_rptr_addr_hi);
data = REG_SET_FIELD(m->sdmax_rlcx_rb_cntl, SDMA0_RLC0_RB_CNTL,
RB_ENABLE, 1);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, data);
return 0;
}
static int kgd_hqd_sdma_dump(struct kgd_dev *kgd,
uint32_t engine_id, uint32_t queue_id,
uint32_t (**dump)[2], uint32_t *n_regs)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint32_t sdma_base_addr = get_sdma_base_addr(adev, engine_id, queue_id);
uint32_t i = 0, reg;
#undef HQD_N_REGS
#define HQD_N_REGS (19+6+7+10)
pr_debug("sdma dump engine id %d queue_id %d\n", engine_id, queue_id);
pr_debug("sdma base addr %x\n", sdma_base_addr);
*dump = kmalloc(HQD_N_REGS*2*sizeof(uint32_t), GFP_KERNEL);
if (*dump == NULL)
return -ENOMEM;
for (reg = mmSDMA0_RLC0_RB_CNTL; reg <= mmSDMA0_RLC0_DOORBELL; reg++)
DUMP_REG(sdma_base_addr + reg);
for (reg = mmSDMA0_RLC0_STATUS; reg <= mmSDMA0_RLC0_CSA_ADDR_HI; reg++)
DUMP_REG(sdma_base_addr + reg);
for (reg = mmSDMA0_RLC0_IB_SUB_REMAIN;
reg <= mmSDMA0_RLC0_MINOR_PTR_UPDATE; reg++)
DUMP_REG(sdma_base_addr + reg);
for (reg = mmSDMA0_RLC0_MIDCMD_DATA0;
reg <= mmSDMA0_RLC0_MIDCMD_CNTL; reg++)
DUMP_REG(sdma_base_addr + reg);
WARN_ON_ONCE(i != HQD_N_REGS);
*n_regs = i;
return 0;
}
static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
uint32_t pipe_id, uint32_t queue_id)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint32_t act;
bool retval = false;
uint32_t low, high;
acquire_queue(kgd, pipe_id, queue_id);
act = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_ACTIVE));
if (act) {
low = lower_32_bits(queue_address >> 8);
high = upper_32_bits(queue_address >> 8);
if (low == RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_BASE)) &&
high == RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_PQ_BASE_HI)))
retval = true;
}
release_queue(kgd);
return retval;
}
static bool kgd_hqd_sdma_is_occupied(struct kgd_dev *kgd, void *mqd)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
struct v10_sdma_mqd *m;
uint32_t sdma_base_addr;
uint32_t sdma_rlc_rb_cntl;
m = get_sdma_mqd(mqd);
sdma_base_addr = get_sdma_base_addr(adev, m->sdma_engine_id,
m->sdma_queue_id);
sdma_rlc_rb_cntl = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
if (sdma_rlc_rb_cntl & SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK)
return true;
return false;
}
static int kgd_hqd_destroy(struct kgd_dev *kgd, void *mqd,
enum kfd_preempt_type reset_type,
unsigned int utimeout, uint32_t pipe_id,
uint32_t queue_id)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
enum hqd_dequeue_request_type type;
unsigned long end_jiffies;
uint32_t temp;
struct v10_compute_mqd *m = get_mqd(mqd);
#if 0
unsigned long flags;
int retry;
#endif
acquire_queue(kgd, pipe_id, queue_id);
if (m->cp_hqd_vmid == 0)
WREG32_FIELD15(GC, 0, RLC_CP_SCHEDULERS, scheduler1, 0);
switch (reset_type) {
case KFD_PREEMPT_TYPE_WAVEFRONT_DRAIN:
type = DRAIN_PIPE;
break;
case KFD_PREEMPT_TYPE_WAVEFRONT_RESET:
type = RESET_WAVES;
break;
default:
type = DRAIN_PIPE;
break;
}
#if 0 /* Is this still needed? */
/* Workaround: If IQ timer is active and the wait time is close to or
* equal to 0, dequeueing is not safe. Wait until either the wait time
* is larger or timer is cleared. Also, ensure that IQ_REQ_PEND is
* cleared before continuing. Also, ensure wait times are set to at
* least 0x3.
*/
local_irq_save(flags);
preempt_disable();
retry = 5000; /* wait for 500 usecs at maximum */
while (true) {
temp = RREG32(mmCP_HQD_IQ_TIMER);
if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, PROCESSING_IQ)) {
pr_debug("HW is processing IQ\n");
goto loop;
}
if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, ACTIVE)) {
if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, RETRY_TYPE)
== 3) /* SEM-rearm is safe */
break;
/* Wait time 3 is safe for CP, but our MMIO read/write
* time is close to 1 microsecond, so check for 10 to
* leave more buffer room
*/
if (REG_GET_FIELD(temp, CP_HQD_IQ_TIMER, WAIT_TIME)
>= 10)
break;
pr_debug("IQ timer is active\n");
} else
break;
loop:
if (!retry) {
pr_err("CP HQD IQ timer status time out\n");
break;
}
ndelay(100);
--retry;
}
retry = 1000;
while (true) {
temp = RREG32(mmCP_HQD_DEQUEUE_REQUEST);
if (!(temp & CP_HQD_DEQUEUE_REQUEST__IQ_REQ_PEND_MASK))
break;
pr_debug("Dequeue request is pending\n");
if (!retry) {
pr_err("CP HQD dequeue request time out\n");
break;
}
ndelay(100);
--retry;
}
local_irq_restore(flags);
preempt_enable();
#endif
WREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_DEQUEUE_REQUEST), type);
end_jiffies = (utimeout * HZ / 1000) + jiffies;
while (true) {
temp = RREG32(SOC15_REG_OFFSET(GC, 0, mmCP_HQD_ACTIVE));
if (!(temp & CP_HQD_ACTIVE__ACTIVE_MASK))
break;
if (time_after(jiffies, end_jiffies)) {
pr_err("cp queue preemption time out.\n");
release_queue(kgd);
return -ETIME;
}
usleep_range(500, 1000);
}
release_queue(kgd);
return 0;
}
static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void *mqd,
unsigned int utimeout)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
struct v10_sdma_mqd *m;
uint32_t sdma_base_addr;
uint32_t temp;
unsigned long end_jiffies = (utimeout * HZ / 1000) + jiffies;
m = get_sdma_mqd(mqd);
sdma_base_addr = get_sdma_base_addr(adev, m->sdma_engine_id,
m->sdma_queue_id);
temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL);
temp = temp & ~SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK;
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL, temp);
while (true) {
temp = RREG32(sdma_base_addr + mmSDMA0_RLC0_CONTEXT_STATUS);
if (temp & SDMA0_RLC0_CONTEXT_STATUS__IDLE_MASK)
break;
if (time_after(jiffies, end_jiffies))
return -ETIME;
usleep_range(500, 1000);
}
WREG32(sdma_base_addr + mmSDMA0_RLC0_DOORBELL, 0);
WREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL,
RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_CNTL) |
SDMA0_RLC0_RB_CNTL__RB_ENABLE_MASK);
m->sdmax_rlcx_rb_rptr = RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR);
m->sdmax_rlcx_rb_rptr_hi =
RREG32(sdma_base_addr + mmSDMA0_RLC0_RB_RPTR_HI);
return 0;
}
static bool get_atc_vmid_pasid_mapping_valid(struct kgd_dev *kgd,
uint8_t vmid)
{
uint32_t reg;
struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
reg = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+ vmid);
return reg & ATC_VMID0_PASID_MAPPING__VALID_MASK;
}
static uint16_t get_atc_vmid_pasid_mapping_pasid(struct kgd_dev *kgd,
uint8_t vmid)
{
uint32_t reg;
struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
reg = RREG32(SOC15_REG_OFFSET(ATHUB, 0, mmATC_VMID0_PASID_MAPPING)
+ vmid);
return reg & ATC_VMID0_PASID_MAPPING__PASID_MASK;
}
static void write_vmid_invalidate_request(struct kgd_dev *kgd, uint8_t vmid)
{
struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
uint32_t req = (1 << vmid) |
(0 << GCVM_INVALIDATE_ENG0_REQ__FLUSH_TYPE__SHIFT) |/* legacy */
GCVM_INVALIDATE_ENG0_REQ__INVALIDATE_L2_PTES_MASK |
GCVM_INVALIDATE_ENG0_REQ__INVALIDATE_L2_PDE0_MASK |
GCVM_INVALIDATE_ENG0_REQ__INVALIDATE_L2_PDE1_MASK |
GCVM_INVALIDATE_ENG0_REQ__INVALIDATE_L2_PDE2_MASK |
GCVM_INVALIDATE_ENG0_REQ__INVALIDATE_L1_PTES_MASK;
mutex_lock(&adev->srbm_mutex);
/* Use light weight invalidation.
*
* TODO 1: agree on the right set of invalidation registers for
* KFD use. Use the last one for now. Invalidate only GCHUB as
* SDMA is now moved to GCHUB
*
* TODO 2: support range-based invalidation, requires kfg2kgd
* interface change
*/
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32),
0xffffffff);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32),
0x0000001f);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_REQ), req);
while (!(RREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ACK)) &
(1 << vmid)))
cpu_relax();
mutex_unlock(&adev->srbm_mutex);
}
static int invalidate_tlbs_with_kiq(struct amdgpu_device *adev, uint16_t pasid)
{
signed long r;
uint32_t seq;
struct amdgpu_ring *ring = &adev->gfx.kiq.ring;
spin_lock(&adev->gfx.kiq.ring_lock);
amdgpu_ring_alloc(ring, 12); /* fence + invalidate_tlbs package*/
amdgpu_ring_write(ring, PACKET3(PACKET3_INVALIDATE_TLBS, 0));
amdgpu_ring_write(ring,
PACKET3_INVALIDATE_TLBS_DST_SEL(1) |
PACKET3_INVALIDATE_TLBS_PASID(pasid));
amdgpu_fence_emit_polling(ring, &seq);
amdgpu_ring_commit(ring);
spin_unlock(&adev->gfx.kiq.ring_lock);
r = amdgpu_fence_wait_polling(ring, seq, adev->usec_timeout);
if (r < 1) {
DRM_ERROR("wait for kiq fence error: %ld.\n", r);
return -ETIME;
}
return 0;
}
static int invalidate_tlbs(struct kgd_dev *kgd, uint16_t pasid)
{
struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
int vmid;
struct amdgpu_ring *ring = &adev->gfx.kiq.ring;
if (amdgpu_emu_mode == 0 && ring->sched.ready)
return invalidate_tlbs_with_kiq(adev, pasid);
for (vmid = 0; vmid < 16; vmid++) {
if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid))
continue;
if (get_atc_vmid_pasid_mapping_valid(kgd, vmid)) {
if (get_atc_vmid_pasid_mapping_pasid(kgd, vmid)
== pasid) {
write_vmid_invalidate_request(kgd, vmid);
break;
}
}
}
return 0;
}
static int invalidate_tlbs_vmid(struct kgd_dev *kgd, uint16_t vmid)
{
struct amdgpu_device *adev = (struct amdgpu_device *) kgd;
if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
pr_err("non kfd vmid %d\n", vmid);
return 0;
}
write_vmid_invalidate_request(kgd, vmid);
return 0;
}
static int kgd_address_watch_disable(struct kgd_dev *kgd)
{
return 0;
}
static int kgd_address_watch_execute(struct kgd_dev *kgd,
unsigned int watch_point_id,
uint32_t cntl_val,
uint32_t addr_hi,
uint32_t addr_lo)
{
return 0;
}
static int kgd_wave_control_execute(struct kgd_dev *kgd,
uint32_t gfx_index_val,
uint32_t sq_cmd)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint32_t data = 0;
mutex_lock(&adev->grbm_idx_mutex);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX), gfx_index_val);
WREG32(SOC15_REG_OFFSET(GC, 0, mmSQ_CMD), sq_cmd);
data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
INSTANCE_BROADCAST_WRITES, 1);
data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
SA_BROADCAST_WRITES, 1);
data = REG_SET_FIELD(data, GRBM_GFX_INDEX,
SE_BROADCAST_WRITES, 1);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_INDEX), data);
mutex_unlock(&adev->grbm_idx_mutex);
return 0;
}
static uint32_t kgd_address_watch_get_offset(struct kgd_dev *kgd,
unsigned int watch_point_id,
unsigned int reg_offset)
{
return 0;
}
static void set_vm_context_page_table_base(struct kgd_dev *kgd, uint32_t vmid,
uint64_t page_table_base)
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
uint64_t base = page_table_base | AMDGPU_PTE_VALID;
if (!amdgpu_amdkfd_is_kfd_vmid(adev, vmid)) {
pr_err("trying to set page table base for wrong VMID %u\n",
vmid);
return;
}
/* TODO: take advantage of per-process address space size. For
* now, all processes share the same address space size, like
* on GFX8 and older.
*/
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32) + (vmid*2), 0);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32) + (vmid*2), 0);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32) + (vmid*2),
lower_32_bits(adev->vm_manager.max_pfn - 1));
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32) + (vmid*2),
upper_32_bits(adev->vm_manager.max_pfn - 1));
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32) + (vmid*2), lower_32_bits(base));
WREG32(SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32) + (vmid*2), upper_32_bits(base));
}

42
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
Просмотреть файл

@ -1732,37 +1732,19 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info,
ret = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
bo->tbo.ttm->pages);
if (ret) {
bo->tbo.ttm->pages[0] = NULL;
pr_info("%s: Failed to get user pages: %d\n",
pr_debug("%s: Failed to get user pages: %d\n",
__func__, ret);
/* Pretend it succeeded. It will fail later
* with a VM fault if the GPU tries to access
* it. Better than hanging indefinitely with
* stalled user mode queues.
*/
/* Return error -EBUSY or -ENOMEM, retry restore */
return ret;
}
amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
}
return 0;
}
/* Remove invalid userptr BOs from hmm track list
*
* Stop HMM track the userptr update
*/
static void untrack_invalid_user_pages(struct amdkfd_process_info *process_info)
{
struct kgd_mem *mem, *tmp_mem;
struct amdgpu_bo *bo;
list_for_each_entry_safe(mem, tmp_mem,
&process_info->userptr_inval_list,
validate_list.head) {
bo = mem->bo;
amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
}
}
/* Validate invalid userptr BOs
*
* Validates BOs on the userptr_inval_list, and moves them back to the
@ -1842,13 +1824,6 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
list_move_tail(&mem->validate_list.head,
&process_info->userptr_valid_list);
/* Stop HMM track the userptr update. We dont check the return
* value for concurrent CPU page table update because we will
* reschedule the restore worker if process_info->evicted_bos
* is updated.
*/
amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
/* Update mapping. If the BO was not validated
* (because we couldn't get user pages), this will
* clear the page table entries, which will result in
@ -1947,7 +1922,6 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
}
unlock_out:
untrack_invalid_user_pages(process_info);
mutex_unlock(&process_info->lock);
mmput(mm);
put_task_struct(usertask);
@ -2153,12 +2127,16 @@ int amdgpu_amdkfd_add_gws_to_process(void *info, void *gws, struct kgd_mem **mem
* Add process eviction fence to bo so they can
* evict each other.
*/
ret = reservation_object_reserve_shared(gws_bo->tbo.resv, 1);
if (ret)
goto reserve_shared_fail;
amdgpu_bo_fence(gws_bo, &process_info->eviction_fence->base, true);
amdgpu_bo_unreserve(gws_bo);
mutex_unlock(&(*mem)->process_info->lock);
return ret;
reserve_shared_fail:
bo_validation_failure:
amdgpu_bo_unreserve(gws_bo);
bo_reservation_failure:

53
drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
Просмотреть файл

@ -118,6 +118,7 @@ union umc_info {
union vram_info {
struct atom_vram_info_header_v2_3 v23;
struct atom_vram_info_header_v2_4 v24;
};
/*
* Return vram width from integrated system info table, if available,
@ -126,22 +127,50 @@ union vram_info {
int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev)
{
struct amdgpu_mode_info *mode_info = &adev->mode_info;
int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
integratedsysteminfo);
int index;
u16 data_offset, size;
union igp_info *igp_info;
union vram_info *vram_info;
u32 mem_channel_number;
u32 mem_channel_width;
u8 frev, crev;
if (adev->flags & AMD_IS_APU)
index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
integratedsysteminfo);
else
index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
vram_info);
/* get any igp specific overrides */
if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, &size,
&frev, &crev, &data_offset)) {
igp_info = (union igp_info *)
(mode_info->atom_context->bios + data_offset);
switch (crev) {
case 11:
return igp_info->v11.umachannelnumber * 64;
default:
return 0;
if (adev->flags & AMD_IS_APU) {
igp_info = (union igp_info *)
(mode_info->atom_context->bios + data_offset);
switch (crev) {
case 11:
mem_channel_number = igp_info->v11.umachannelnumber;
/* channel width is 64 */
return mem_channel_number * 64;
default:
return 0;
}
} else {
vram_info = (union vram_info *)
(mode_info->atom_context->bios + data_offset);
switch (crev) {
case 3:
mem_channel_number = vram_info->v23.vram_module[0].channel_num;
mem_channel_width = vram_info->v23.vram_module[0].channel_width;
return mem_channel_number * (1 << mem_channel_width);
case 4:
mem_channel_number = vram_info->v24.vram_module[0].channel_num;
mem_channel_width = vram_info->v24.vram_module[0].channel_width;
return mem_channel_number * (1 << mem_channel_width);
default:
return 0;
}
}
}
@ -179,6 +208,9 @@ static int convert_atom_mem_type_to_vram_type (struct amdgpu_device *adev,
case ATOM_DGPU_VRAM_TYPE_HBM2:
vram_type = AMDGPU_VRAM_TYPE_HBM;
break;
case ATOM_DGPU_VRAM_TYPE_GDDR6:
vram_type = AMDGPU_VRAM_TYPE_GDDR6;
break;
default:
vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
break;
@ -227,6 +259,9 @@ int amdgpu_atomfirmware_get_vram_type(struct amdgpu_device *adev)
case 3:
mem_type = vram_info->v23.vram_module[0].memory_type;
return convert_atom_mem_type_to_vram_type(adev, mem_type);
case 4:
mem_type = vram_info->v24.vram_module[0].memory_type;
return convert_atom_mem_type_to_vram_type(adev, mem_type);
default:
return 0;
}

17
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
Просмотреть файл

@ -650,7 +650,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
}
r = ttm_eu_reserve_buffers(&p->ticket, &p->validated, true,
&duplicates, true);
&duplicates, false);
if (unlikely(r != 0)) {
if (r != -ERESTARTSYS)
DRM_ERROR("ttm_eu_reserve_buffers failed.\n");
@ -673,16 +673,12 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
}
r = amdgpu_cs_list_validate(p, &duplicates);
if (r) {
DRM_ERROR("amdgpu_cs_list_validate(duplicates) failed.\n");
if (r)
goto error_validate;
}
r = amdgpu_cs_list_validate(p, &p->validated);
if (r) {
DRM_ERROR("amdgpu_cs_list_validate(validated) failed.\n");
if (r)
goto error_validate;
}
amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
p->bytes_moved_vis);
@ -878,7 +874,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
if (r)
return r;
if (amdgpu_sriov_vf(adev)) {
if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
struct dma_fence *f;
bo_va = fpriv->csa_va;
@ -967,7 +963,8 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
continue;
if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX && amdgpu_sriov_vf(adev)) {
if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX &&
(amdgpu_mcbp || amdgpu_sriov_vf(adev))) {
if (chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
ce_preempt++;
@ -1385,7 +1382,7 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
if (r) {
if (r == -ENOMEM)
DRM_ERROR("Not enough memory for command submission!\n");
else if (r != -ERESTARTSYS)
else if (r != -ERESTARTSYS && r != -EAGAIN)
DRM_ERROR("Failed to process the buffer list %d!\n", r);
goto out;
}

3
drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
Просмотреть файл

@ -47,6 +47,7 @@ int amdgpu_allocate_static_csa(struct amdgpu_device *adev, struct amdgpu_bo **bo
return -ENOMEM;
memset(ptr, 0, size);
adev->virt.csa_cpu_addr = ptr;
return 0;
}
@ -79,7 +80,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
list_add(&csa_tv.head, &list);
amdgpu_vm_get_pd_bo(vm, &list, &pd);
r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, true);
r = ttm_eu_reserve_buffers(&ticket, &list, true, NULL, false);
if (r) {
DRM_ERROR("failed to reserve CSA,PD BOs: err=%d\n", r);
return r;

178
drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
Просмотреть файл

@ -923,17 +923,195 @@ static const struct drm_info_list amdgpu_debugfs_list[] = {
{"amdgpu_evict_gtt", &amdgpu_debugfs_evict_gtt},
};
static void amdgpu_ib_preempt_fences_swap(struct amdgpu_ring *ring,
struct dma_fence **fences)
{
struct amdgpu_fence_driver *drv = &ring->fence_drv;
uint32_t sync_seq, last_seq;
last_seq = atomic_read(&ring->fence_drv.last_seq);
sync_seq = ring->fence_drv.sync_seq;
last_seq &= drv->num_fences_mask;
sync_seq &= drv->num_fences_mask;
do {
struct dma_fence *fence, **ptr;
++last_seq;
last_seq &= drv->num_fences_mask;
ptr = &drv->fences[last_seq];
fence = rcu_dereference_protected(*ptr, 1);
RCU_INIT_POINTER(*ptr, NULL);
if (!fence)
continue;
fences[last_seq] = fence;
} while (last_seq != sync_seq);
}
static void amdgpu_ib_preempt_signal_fences(struct dma_fence **fences,
int length)
{
int i;
struct dma_fence *fence;
for (i = 0; i < length; i++) {
fence = fences[i];
if (!fence)
continue;
dma_fence_signal(fence);
dma_fence_put(fence);
}
}
static void amdgpu_ib_preempt_job_recovery(struct drm_gpu_scheduler *sched)
{
struct drm_sched_job *s_job;
struct dma_fence *fence;
spin_lock(&sched->job_list_lock);
list_for_each_entry(s_job, &sched->ring_mirror_list, node) {
fence = sched->ops->run_job(s_job);
dma_fence_put(fence);
}
spin_unlock(&sched->job_list_lock);
}
static void amdgpu_ib_preempt_mark_partial_job(struct amdgpu_ring *ring)
{
struct amdgpu_job *job;
struct drm_sched_job *s_job;
uint32_t preempt_seq;
struct dma_fence *fence, **ptr;
struct amdgpu_fence_driver *drv = &ring->fence_drv;
struct drm_gpu_scheduler *sched = &ring->sched;
if (ring->funcs->type != AMDGPU_RING_TYPE_GFX)
return;
preempt_seq = le32_to_cpu(*(drv->cpu_addr + 2));
if (preempt_seq <= atomic_read(&drv->last_seq))
return;
preempt_seq &= drv->num_fences_mask;
ptr = &drv->fences[preempt_seq];
fence = rcu_dereference_protected(*ptr, 1);
spin_lock(&sched->job_list_lock);
list_for_each_entry(s_job, &sched->ring_mirror_list, node) {
job = to_amdgpu_job(s_job);
if (job->fence == fence)
/* mark the job as preempted */
job->preemption_status |= AMDGPU_IB_PREEMPTED;
}
spin_unlock(&sched->job_list_lock);
}
static int amdgpu_debugfs_ib_preempt(void *data, u64 val)
{
int r, resched, length;
struct amdgpu_ring *ring;
struct dma_fence **fences = NULL;
struct amdgpu_device *adev = (struct amdgpu_device *)data;
if (val >= AMDGPU_MAX_RINGS)
return -EINVAL;
ring = adev->rings[val];
if (!ring || !ring->funcs->preempt_ib || !ring->sched.thread)
return -EINVAL;
/* the last preemption failed */
if (ring->trail_seq != le32_to_cpu(*ring->trail_fence_cpu_addr))
return -EBUSY;
length = ring->fence_drv.num_fences_mask + 1;
fences = kcalloc(length, sizeof(void *), GFP_KERNEL);
if (!fences)
return -ENOMEM;
/* stop the scheduler */
kthread_park(ring->sched.thread);
resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
/* preempt the IB */
r = amdgpu_ring_preempt_ib(ring);
if (r) {
DRM_WARN("failed to preempt ring %d\n", ring->idx);
goto failure;
}
amdgpu_fence_process(ring);
if (atomic_read(&ring->fence_drv.last_seq) !=
ring->fence_drv.sync_seq) {
DRM_INFO("ring %d was preempted\n", ring->idx);
amdgpu_ib_preempt_mark_partial_job(ring);
/* swap out the old fences */
amdgpu_ib_preempt_fences_swap(ring, fences);
amdgpu_fence_driver_force_completion(ring);
/* resubmit unfinished jobs */
amdgpu_ib_preempt_job_recovery(&ring->sched);
/* wait for jobs finished */
amdgpu_fence_wait_empty(ring);
/* signal the old fences */
amdgpu_ib_preempt_signal_fences(fences, length);
}
failure:
/* restart the scheduler */
kthread_unpark(ring->sched.thread);
ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
if (fences)
kfree(fences);
return 0;
}
DEFINE_SIMPLE_ATTRIBUTE(fops_ib_preempt, NULL,
amdgpu_debugfs_ib_preempt, "%llu\n");
int amdgpu_debugfs_init(struct amdgpu_device *adev)
{
adev->debugfs_preempt =
debugfs_create_file("amdgpu_preempt_ib", 0600,
adev->ddev->primary->debugfs_root,
(void *)adev, &fops_ib_preempt);
if (!(adev->debugfs_preempt)) {
DRM_ERROR("unable to create amdgpu_preempt_ib debugsfs file\n");
return -EIO;
}
return amdgpu_debugfs_add_files(adev, amdgpu_debugfs_list,
ARRAY_SIZE(amdgpu_debugfs_list));
}
void amdgpu_debugfs_preempt_cleanup(struct amdgpu_device *adev)
{
if (adev->debugfs_preempt)
debugfs_remove(adev->debugfs_preempt);
}
#else
int amdgpu_debugfs_init(struct amdgpu_device *adev)
{
return 0;
}
void amdgpu_debugfs_preempt_cleanup(struct amdgpu_device *adev) { }
int amdgpu_debugfs_regs_init(struct amdgpu_device *adev)
{
return 0;

1
drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.h
Просмотреть файл

@ -34,6 +34,7 @@ struct amdgpu_debugfs {
int amdgpu_debugfs_regs_init(struct amdgpu_device *adev);
void amdgpu_debugfs_regs_cleanup(struct amdgpu_device *adev);
int amdgpu_debugfs_init(struct amdgpu_device *adev);
void amdgpu_debugfs_preempt_cleanup(struct amdgpu_device *adev);
int amdgpu_debugfs_add_files(struct amdgpu_device *adev,
const struct drm_info_list *files,
unsigned nfiles);

100
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
Просмотреть файл

@ -52,6 +52,7 @@
#endif
#include "vi.h"
#include "soc15.h"
#include "nv.h"
#include "bif/bif_4_1_d.h"
#include <linux/pci.h>
#include <linux/firmware.h>
@ -62,12 +63,14 @@
#include "amdgpu_xgmi.h"
#include "amdgpu_ras.h"
#include "amdgpu_pmu.h"
MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/vega12_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/picasso_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/raven2_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/navi10_gpu_info.bin");
#define AMDGPU_RESUME_MS 2000
@ -95,6 +98,7 @@ static const char *amdgpu_asic_name[] = {
"VEGA12",
"VEGA20",
"RAVEN",
"NAVI10",
"LAST",
};
@ -507,7 +511,10 @@ void amdgpu_device_program_register_sequence(struct amdgpu_device *adev,
} else {
tmp = RREG32(reg);
tmp &= ~and_mask;
tmp |= or_mask;
if (adev->family >= AMDGPU_FAMILY_AI)
tmp |= (or_mask & and_mask);
else
tmp |= or_mask;
}
WREG32(reg, tmp);
}
@ -974,13 +981,6 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev)
amdgpu_device_check_block_size(adev);
if (amdgpu_vram_page_split != -1 && (amdgpu_vram_page_split < 16 ||
!is_power_of_2(amdgpu_vram_page_split))) {
dev_warn(adev->dev, "invalid VRAM page split (%d)\n",
amdgpu_vram_page_split);
amdgpu_vram_page_split = 1024;
}
ret = amdgpu_device_get_job_timeout_settings(adev);
if (ret) {
dev_err(adev->dev, "invalid lockup_timeout parameter syntax\n");
@ -1384,6 +1384,9 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
else
chip_name = "raven";
break;
case CHIP_NAVI10:
chip_name = "navi10";
break;
}
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name);
@ -1430,6 +1433,23 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
adev->gfx.cu_info.max_scratch_slots_per_cu =
le32_to_cpu(gpu_info_fw->gc_max_scratch_slots_per_cu);
adev->gfx.cu_info.lds_size = le32_to_cpu(gpu_info_fw->gc_lds_size);
if (hdr->version_minor >= 1) {
const struct gpu_info_firmware_v1_1 *gpu_info_fw =
(const struct gpu_info_firmware_v1_1 *)(adev->firmware.gpu_info_fw->data +
le32_to_cpu(hdr->header.ucode_array_offset_bytes));
adev->gfx.config.num_sc_per_sh =
le32_to_cpu(gpu_info_fw->num_sc_per_sh);
adev->gfx.config.num_packer_per_sc =
le32_to_cpu(gpu_info_fw->num_packer_per_sc);
}
#ifdef CONFIG_DRM_AMD_DC_DCN2_0
if (hdr->version_minor == 2) {
const struct gpu_info_firmware_v1_2 *gpu_info_fw =
(const struct gpu_info_firmware_v1_2 *)(adev->firmware.gpu_info_fw->data +
le32_to_cpu(hdr->header.ucode_array_offset_bytes));
adev->dm.soc_bounding_box = &gpu_info_fw->soc_bounding_box;
}
#endif
break;
}
default:
@ -1518,6 +1538,13 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
if (r)
return r;
break;
case CHIP_NAVI10:
adev->family = AMDGPU_FAMILY_NV;
r = nv_set_ip_blocks(adev);
if (r)
return r;
break;
default:
/* FIXME: not supported yet */
return -EINVAL;
@ -1542,17 +1569,6 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
if (amdgpu_sriov_vf(adev))
adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
/* Read BIOS */
if (!amdgpu_get_bios(adev))
return -EINVAL;
r = amdgpu_atombios_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_atombios_init failed\n");
amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0);
return r;
}
for (i = 0; i < adev->num_ip_blocks; i++) {
if ((amdgpu_ip_block_mask & (1 << i)) == 0) {
DRM_ERROR("disabled ip block: %d <%s>\n",
@ -1574,6 +1590,19 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
adev->ip_blocks[i].status.valid = true;
}
}
/* get the vbios after the asic_funcs are set up */
if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON) {
/* Read BIOS */
if (!amdgpu_get_bios(adev))
return -EINVAL;
r = amdgpu_atombios_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_atombios_init failed\n");
amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0);
return r;
}
}
}
adev->cg_flags &= amdgpu_cg_mask;
@ -1713,7 +1742,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
adev->ip_blocks[i].status.hw = true;
/* right after GMC hw init, we create CSA */
if (amdgpu_sriov_vf(adev)) {
if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
r = amdgpu_allocate_static_csa(adev, &adev->virt.csa_obj,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_CSA_SIZE);
@ -2394,6 +2423,9 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type)
case CHIP_VEGA20:
#if defined(CONFIG_DRM_AMD_DC_DCN1_0)
case CHIP_RAVEN:
#endif
#if defined(CONFIG_DRM_AMD_DC_DCN2_0)
case CHIP_NAVI10:
#endif
return amdgpu_dc != 0;
#endif
@ -2566,6 +2598,20 @@ int amdgpu_device_init(struct amdgpu_device *adev,
amdgpu_device_get_pcie_info(adev);
if (amdgpu_mcbp)
DRM_INFO("MCBP is enabled\n");
if (amdgpu_mes && adev->asic_type >= CHIP_NAVI10)
adev->enable_mes = true;
if (amdgpu_discovery) {
r = amdgpu_discovery_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_discovery_init failed\n");
return r;
}
}
/* early init functions */
r = amdgpu_device_ip_early_init(adev);
if (r)
@ -2690,6 +2736,9 @@ fence_driver_init:
amdgpu_fbdev_init(adev);
if (amdgpu_sriov_vf(adev) && amdgim_is_hwperf(adev))
amdgpu_pm_virt_sysfs_init(adev);
r = amdgpu_pm_sysfs_init(adev);
if (r)
DRM_ERROR("registering pm debugfs failed (%d).\n", r);
@ -2749,6 +2798,10 @@ fence_driver_init:
return r;
}
r = amdgpu_pmu_init(adev);
if (r)
dev_err(adev->dev, "amdgpu_pmu_init failed\n");
return 0;
failed:
@ -2811,9 +2864,16 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
iounmap(adev->rmmio);
adev->rmmio = NULL;
amdgpu_device_doorbell_fini(adev);
if (amdgpu_sriov_vf(adev) && amdgim_is_hwperf(adev))
amdgpu_pm_virt_sysfs_fini(adev);
amdgpu_debugfs_regs_cleanup(adev);
device_remove_file(adev->dev, &dev_attr_pcie_replay_count);
amdgpu_ucode_sysfs_fini(adev);
amdgpu_pmu_fini(adev);
amdgpu_debugfs_preempt_cleanup(adev);
if (amdgpu_discovery)
amdgpu_discovery_fini(adev);
}

415
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c Normal file
Просмотреть файл

@ -0,0 +1,415 @@
/*
* Copyright 2018 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.
*
*/
#include "amdgpu.h"
#include "amdgpu_discovery.h"
#include "soc15_common.h"
#include "soc15_hw_ip.h"
#include "nbio/nbio_2_3_offset.h"
#include "discovery.h"
#define mmRCC_CONFIG_MEMSIZE 0xde3
#define mmMM_INDEX 0x0
#define mmMM_INDEX_HI 0x6
#define mmMM_DATA 0x1
#define HW_ID_MAX 300
const char *hw_id_names[HW_ID_MAX] = {
[MP1_HWID] = "MP1",
[MP2_HWID] = "MP2",
[THM_HWID] = "THM",
[SMUIO_HWID] = "SMUIO",
[FUSE_HWID] = "FUSE",
[CLKA_HWID] = "CLKA",
[PWR_HWID] = "PWR",
[GC_HWID] = "GC",
[UVD_HWID] = "UVD",
[AUDIO_AZ_HWID] = "AUDIO_AZ",
[ACP_HWID] = "ACP",
[DCI_HWID] = "DCI",
[DMU_HWID] = "DMU",
[DCO_HWID] = "DCO",
[DIO_HWID] = "DIO",
[XDMA_HWID] = "XDMA",
[DCEAZ_HWID] = "DCEAZ",
[DAZ_HWID] = "DAZ",
[SDPMUX_HWID] = "SDPMUX",
[NTB_HWID] = "NTB",
[IOHC_HWID] = "IOHC",
[L2IMU_HWID] = "L2IMU",
[VCE_HWID] = "VCE",
[MMHUB_HWID] = "MMHUB",
[ATHUB_HWID] = "ATHUB",
[DBGU_NBIO_HWID] = "DBGU_NBIO",
[DFX_HWID] = "DFX",
[DBGU0_HWID] = "DBGU0",
[DBGU1_HWID] = "DBGU1",
[OSSSYS_HWID] = "OSSSYS",
[HDP_HWID] = "HDP",
[SDMA0_HWID] = "SDMA0",
[SDMA1_HWID] = "SDMA1",
[ISP_HWID] = "ISP",
[DBGU_IO_HWID] = "DBGU_IO",
[DF_HWID] = "DF",
[CLKB_HWID] = "CLKB",
[FCH_HWID] = "FCH",
[DFX_DAP_HWID] = "DFX_DAP",
[L1IMU_PCIE_HWID] = "L1IMU_PCIE",
[L1IMU_NBIF_HWID] = "L1IMU_NBIF",
[L1IMU_IOAGR_HWID] = "L1IMU_IOAGR",
[L1IMU3_HWID] = "L1IMU3",
[L1IMU4_HWID] = "L1IMU4",
[L1IMU5_HWID] = "L1IMU5",
[L1IMU6_HWID] = "L1IMU6",
[L1IMU7_HWID] = "L1IMU7",
[L1IMU8_HWID] = "L1IMU8",
[L1IMU9_HWID] = "L1IMU9",
[L1IMU10_HWID] = "L1IMU10",
[L1IMU11_HWID] = "L1IMU11",
[L1IMU12_HWID] = "L1IMU12",
[L1IMU13_HWID] = "L1IMU13",
[L1IMU14_HWID] = "L1IMU14",
[L1IMU15_HWID] = "L1IMU15",
[WAFLC_HWID] = "WAFLC",
[FCH_USB_PD_HWID] = "FCH_USB_PD",
[PCIE_HWID] = "PCIE",
[PCS_HWID] = "PCS",
[DDCL_HWID] = "DDCL",
[SST_HWID] = "SST",
[IOAGR_HWID] = "IOAGR",
[NBIF_HWID] = "NBIF",
[IOAPIC_HWID] = "IOAPIC",
[SYSTEMHUB_HWID] = "SYSTEMHUB",
[NTBCCP_HWID] = "NTBCCP",
[UMC_HWID] = "UMC",
[SATA_HWID] = "SATA",
[USB_HWID] = "USB",
[CCXSEC_HWID] = "CCXSEC",
[XGMI_HWID] = "XGMI",
[XGBE_HWID] = "XGBE",
[MP0_HWID] = "MP0",
};
static int hw_id_map[MAX_HWIP] = {
[GC_HWIP] = GC_HWID,
[HDP_HWIP] = HDP_HWID,
[SDMA0_HWIP] = SDMA0_HWID,
[SDMA1_HWIP] = SDMA1_HWID,
[MMHUB_HWIP] = MMHUB_HWID,
[ATHUB_HWIP] = ATHUB_HWID,
[NBIO_HWIP] = NBIF_HWID,
[MP0_HWIP] = MP0_HWID,
[MP1_HWIP] = MP1_HWID,
[UVD_HWIP] = UVD_HWID,
[VCE_HWIP] = VCE_HWID,
[DF_HWIP] = DF_HWID,
[DCE_HWIP] = DCEAZ_HWID,
[OSSSYS_HWIP] = OSSSYS_HWID,
[SMUIO_HWIP] = SMUIO_HWID,
[PWR_HWIP] = PWR_HWID,
[NBIF_HWIP] = NBIF_HWID,
[THM_HWIP] = THM_HWID,
[CLK_HWIP] = CLKA_HWID,
};
static int amdgpu_discovery_read_binary(struct amdgpu_device *adev, uint8_t *binary)
{
uint32_t *p = (uint32_t *)binary;
uint64_t vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20;
uint64_t pos = vram_size - BINARY_MAX_SIZE;
unsigned long flags;
while (pos < vram_size) {
spin_lock_irqsave(&adev->mmio_idx_lock, flags);
WREG32_NO_KIQ(mmMM_INDEX, ((uint32_t)pos) | 0x80000000);
WREG32_NO_KIQ(mmMM_INDEX_HI, pos >> 31);
*p++ = RREG32_NO_KIQ(mmMM_DATA);
spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
pos += 4;
}
return 0;
}
static uint16_t amdgpu_discovery_calculate_checksum(uint8_t *data, uint32_t size)
{
uint16_t checksum = 0;
int i;
for (i = 0; i < size; i++)
checksum += data[i];
return checksum;
}
static inline bool amdgpu_discovery_verify_checksum(uint8_t *data, uint32_t size,
uint16_t expected)
{
return !!(amdgpu_discovery_calculate_checksum(data, size) == expected);
}
int amdgpu_discovery_init(struct amdgpu_device *adev)
{
struct table_info *info;
struct binary_header *bhdr;
struct ip_discovery_header *ihdr;
struct gpu_info_header *ghdr;
uint16_t offset;
uint16_t size;
uint16_t checksum;
int r;
adev->discovery = kzalloc(BINARY_MAX_SIZE, GFP_KERNEL);
if (!adev->discovery)
return -ENOMEM;
r = amdgpu_discovery_read_binary(adev, adev->discovery);
if (r) {
DRM_ERROR("failed to read ip discovery binary\n");
goto out;
}
bhdr = (struct binary_header *)adev->discovery;
if (le32_to_cpu(bhdr->binary_signature) != BINARY_SIGNATURE) {
DRM_ERROR("invalid ip discovery binary signature\n");
r = -EINVAL;
goto out;
}
offset = offsetof(struct binary_header, binary_checksum) +
sizeof(bhdr->binary_checksum);
size = bhdr->binary_size - offset;
checksum = bhdr->binary_checksum;
if (!amdgpu_discovery_verify_checksum(adev->discovery + offset,
size, checksum)) {
DRM_ERROR("invalid ip discovery binary checksum\n");
r = -EINVAL;
goto out;
}
info = &bhdr->table_list[IP_DISCOVERY];
offset = le16_to_cpu(info->offset);
checksum = le16_to_cpu(info->checksum);
ihdr = (struct ip_discovery_header *)(adev->discovery + offset);
if (le32_to_cpu(ihdr->signature) != DISCOVERY_TABLE_SIGNATURE) {
DRM_ERROR("invalid ip discovery data table signature\n");
r = -EINVAL;
goto out;
}
if (!amdgpu_discovery_verify_checksum(adev->discovery + offset,
ihdr->size, checksum)) {
DRM_ERROR("invalid ip discovery data table checksum\n");
r = -EINVAL;
goto out;
}
info = &bhdr->table_list[GC];
offset = le16_to_cpu(info->offset);
checksum = le16_to_cpu(info->checksum);
ghdr = (struct gpu_info_header *)(adev->discovery + offset);
if (!amdgpu_discovery_verify_checksum(adev->discovery + offset,
ghdr->size, checksum)) {
DRM_ERROR("invalid gc data table checksum\n");
r = -EINVAL;
goto out;
}
return 0;
out:
kfree(adev->discovery);
adev->discovery = NULL;
return r;
}
void amdgpu_discovery_fini(struct amdgpu_device *adev)
{
kfree(adev->discovery);
adev->discovery = NULL;
}
int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)
{
struct binary_header *bhdr;
struct ip_discovery_header *ihdr;
struct die_header *dhdr;
struct ip *ip;
uint16_t die_offset;
uint16_t ip_offset;
uint16_t num_dies;
uint16_t num_ips;
uint8_t num_base_address;
int hw_ip;
int i, j, k;
if (!adev->discovery) {
DRM_ERROR("ip discovery uninitialized\n");
return -EINVAL;
}
bhdr = (struct binary_header *)adev->discovery;
ihdr = (struct ip_discovery_header *)(adev->discovery +
le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
num_dies = le16_to_cpu(ihdr->num_dies);
DRM_DEBUG("number of dies: %d\n", num_dies);
for (i = 0; i < num_dies; i++) {
die_offset = le16_to_cpu(ihdr->die_info[i].die_offset);
dhdr = (struct die_header *)(adev->discovery + die_offset);
num_ips = le16_to_cpu(dhdr->num_ips);
ip_offset = die_offset + sizeof(*dhdr);
if (le16_to_cpu(dhdr->die_id) != i) {
DRM_ERROR("invalid die id %d, expected %d\n",
le16_to_cpu(dhdr->die_id), i);
return -EINVAL;
}
DRM_DEBUG("number of hardware IPs on die%d: %d\n",
le16_to_cpu(dhdr->die_id), num_ips);
for (j = 0; j < num_ips; j++) {
ip = (struct ip *)(adev->discovery + ip_offset);
num_base_address = ip->num_base_address;
DRM_DEBUG("%s(%d) #%d v%d.%d.%d:\n",
hw_id_names[le16_to_cpu(ip->hw_id)],
le16_to_cpu(ip->hw_id),
ip->number_instance,
ip->major, ip->minor,
ip->revision);
for (k = 0; k < num_base_address; k++) {
/*
* convert the endianness of base addresses in place,
* so that we don't need to convert them when accessing adev->reg_offset.
*/
ip->base_address[k] = le32_to_cpu(ip->base_address[k]);
DRM_DEBUG("\t0x%08x\n", ip->base_address[k]);
}
for (hw_ip = 0; hw_ip < MAX_HWIP; hw_ip++) {
if (hw_id_map[hw_ip] == le16_to_cpu(ip->hw_id)) {
DRM_INFO("set register base offset for %s\n",
hw_id_names[le16_to_cpu(ip->hw_id)]);
adev->reg_offset[hw_ip][ip->number_instance] =
ip->base_address;
}
}
ip_offset += sizeof(*ip) + 4 * (ip->num_base_address - 1);
}
}
return 0;
}
int amdgpu_discovery_get_ip_version(struct amdgpu_device *adev, int hw_id,
int *major, int *minor)
{
struct binary_header *bhdr;
struct ip_discovery_header *ihdr;
struct die_header *dhdr;
struct ip *ip;
uint16_t die_offset;
uint16_t ip_offset;
uint16_t num_dies;
uint16_t num_ips;
int i, j;
if (!adev->discovery) {
DRM_ERROR("ip discovery uninitialized\n");
return -EINVAL;
}
bhdr = (struct binary_header *)adev->discovery;
ihdr = (struct ip_discovery_header *)(adev->discovery +
le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
num_dies = le16_to_cpu(ihdr->num_dies);
for (i = 0; i < num_dies; i++) {
die_offset = le16_to_cpu(ihdr->die_info[i].die_offset);
dhdr = (struct die_header *)(adev->discovery + die_offset);
num_ips = le16_to_cpu(dhdr->num_ips);
ip_offset = die_offset + sizeof(*dhdr);
for (j = 0; j < num_ips; j++) {
ip = (struct ip *)(adev->discovery + ip_offset);
if (le16_to_cpu(ip->hw_id) == hw_id) {
if (major)
*major = ip->major;
if (minor)
*minor = ip->minor;
return 0;
}
ip_offset += sizeof(*ip) + 4 * (ip->num_base_address - 1);
}
}
return -EINVAL;
}
int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev)
{
struct binary_header *bhdr;
struct gc_info_v1_0 *gc_info;
if (!adev->discovery) {
DRM_ERROR("ip discovery uninitialized\n");
return -EINVAL;
}
bhdr = (struct binary_header *)adev->discovery;
gc_info = (struct gc_info_v1_0 *)(adev->discovery +
le16_to_cpu(bhdr->table_list[GC].offset));
adev->gfx.config.max_shader_engines = le32_to_cpu(gc_info->gc_num_se);
adev->gfx.config.max_cu_per_sh = 2 * (le32_to_cpu(gc_info->gc_num_wgp0_per_sa) +
le32_to_cpu(gc_info->gc_num_wgp1_per_sa));
adev->gfx.config.max_sh_per_se = le32_to_cpu(gc_info->gc_num_sa_per_se);
adev->gfx.config.max_backends_per_se = le32_to_cpu(gc_info->gc_num_rb_per_se);
adev->gfx.config.max_texture_channel_caches = le32_to_cpu(gc_info->gc_num_gl2c);
adev->gfx.config.max_gprs = le32_to_cpu(gc_info->gc_num_gprs);
adev->gfx.config.max_gs_threads = le32_to_cpu(gc_info->gc_num_max_gs_thds);
adev->gfx.config.gs_vgt_table_depth = le32_to_cpu(gc_info->gc_gs_table_depth);
adev->gfx.config.gs_prim_buffer_depth = le32_to_cpu(gc_info->gc_gsprim_buff_depth);
adev->gfx.config.double_offchip_lds_buf = le32_to_cpu(gc_info->gc_double_offchip_lds_buffer);
adev->gfx.cu_info.wave_front_size = le32_to_cpu(gc_info->gc_wave_size);
adev->gfx.cu_info.max_waves_per_simd = le32_to_cpu(gc_info->gc_max_waves_per_simd);
adev->gfx.cu_info.max_scratch_slots_per_cu = le32_to_cpu(gc_info->gc_max_scratch_slots_per_cu);
adev->gfx.cu_info.lds_size = le32_to_cpu(gc_info->gc_lds_size);
adev->gfx.config.num_sc_per_sh = le32_to_cpu(gc_info->gc_num_sc_per_se) /
le32_to_cpu(gc_info->gc_num_sa_per_se);
adev->gfx.config.num_packer_per_sc = le32_to_cpu(gc_info->gc_num_packer_per_sc);
return 0;
}

34
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h Normal file
Просмотреть файл

@ -0,0 +1,34 @@
/*
* Copyright 2018 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 __AMDGPU_DISCOVERY__
#define __AMDGPU_DISCOVERY__
int amdgpu_discovery_init(struct amdgpu_device *adev);
void amdgpu_discovery_fini(struct amdgpu_device *adev);
int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev);
int amdgpu_discovery_get_ip_version(struct amdgpu_device *adev, int hw_id,
int *major, int *minor);
int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev);
#endif /* __AMDGPU_DISCOVERY__ */

40
drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
Просмотреть файл

@ -51,6 +51,7 @@ struct amdgpu_doorbell_index {
uint32_t userqueue_start;
uint32_t userqueue_end;
uint32_t gfx_ring0;
uint32_t gfx_ring1;
uint32_t sdma_engine[8];
uint32_t ih;
union {
@ -153,6 +154,45 @@ typedef enum _AMDGPU_VEGA20_DOORBELL_ASSIGNMENT
AMDGPU_VEGA20_DOORBELL_INVALID = 0xFFFF
} AMDGPU_VEGA20_DOORBELL_ASSIGNMENT;
typedef enum _AMDGPU_NAVI10_DOORBELL_ASSIGNMENT
{
/* Compute + GFX: 0~255 */
AMDGPU_NAVI10_DOORBELL_KIQ = 0x000,
AMDGPU_NAVI10_DOORBELL_HIQ = 0x001,
AMDGPU_NAVI10_DOORBELL_DIQ = 0x002,
AMDGPU_NAVI10_DOORBELL_MEC_RING0 = 0x003,
AMDGPU_NAVI10_DOORBELL_MEC_RING1 = 0x004,
AMDGPU_NAVI10_DOORBELL_MEC_RING2 = 0x005,
AMDGPU_NAVI10_DOORBELL_MEC_RING3 = 0x006,
AMDGPU_NAVI10_DOORBELL_MEC_RING4 = 0x007,
AMDGPU_NAVI10_DOORBELL_MEC_RING5 = 0x008,
AMDGPU_NAVI10_DOORBELL_MEC_RING6 = 0x009,
AMDGPU_NAVI10_DOORBELL_MEC_RING7 = 0x00A,
AMDGPU_NAVI10_DOORBELL_USERQUEUE_START = 0x00B,
AMDGPU_NAVI10_DOORBELL_USERQUEUE_END = 0x08A,
AMDGPU_NAVI10_DOORBELL_GFX_RING0 = 0x08B,
AMDGPU_NAVI10_DOORBELL_GFX_RING1 = 0x08C,
/* SDMA:256~335*/
AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0 = 0x100,
AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE1 = 0x10A,
/* IH: 376~391 */
AMDGPU_NAVI10_DOORBELL_IH = 0x178,
/* MMSCH: 392~407
* overlap the doorbell assignment with VCN as they are mutually exclusive
* VCE engine's doorbell is 32 bit and two VCE ring share one QWORD
*/
AMDGPU_NAVI10_DOORBELL64_VCN0_1 = 0x188, /* lower 32 bits for VNC0 and upper 32 bits for VNC1 */
AMDGPU_NAVI10_DOORBELL64_VCN2_3 = 0x189,
AMDGPU_NAVI10_DOORBELL64_VCN4_5 = 0x18A,
AMDGPU_NAVI10_DOORBELL64_VCN6_7 = 0x18B,
AMDGPU_NAVI10_DOORBELL64_FIRST_NON_CP = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0,
AMDGPU_NAVI10_DOORBELL64_LAST_NON_CP = AMDGPU_NAVI10_DOORBELL64_VCN6_7,
AMDGPU_NAVI10_DOORBELL_MAX_ASSIGNMENT = 0x18F,
AMDGPU_NAVI10_DOORBELL_INVALID = 0xFFFF
} AMDGPU_NAVI10_DOORBELL_ASSIGNMENT;
/*
* 64bit doorbell, offset are in QWORD, occupy 2KB doorbell space
*/

59
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.c
Просмотреть файл

@ -906,16 +906,63 @@ amdgpu_get_vce_clock_state(void *handle, u32 idx)
int amdgpu_dpm_get_sclk(struct amdgpu_device *adev, bool low)
{
if (is_support_sw_smu(adev))
return smu_get_sclk(&adev->smu, low);
else
uint32_t clk_freq;
int ret = 0;
if (is_support_sw_smu(adev)) {
ret = smu_get_dpm_freq_range(&adev->smu, SMU_GFXCLK,
low ? &clk_freq : NULL,
!low ? &clk_freq : NULL);
if (ret)
return 0;
return clk_freq * 100;
} else {
return (adev)->powerplay.pp_funcs->get_sclk((adev)->powerplay.pp_handle, (low));
}
}
int amdgpu_dpm_get_mclk(struct amdgpu_device *adev, bool low)
{
if (is_support_sw_smu(adev))
return smu_get_mclk(&adev->smu, low);
else
uint32_t clk_freq;
int ret = 0;
if (is_support_sw_smu(adev)) {
ret = smu_get_dpm_freq_range(&adev->smu, SMU_UCLK,
low ? &clk_freq : NULL,
!low ? &clk_freq : NULL);
if (ret)
return 0;
return clk_freq * 100;
} else {
return (adev)->powerplay.pp_funcs->get_mclk((adev)->powerplay.pp_handle, (low));
}
}
int amdgpu_dpm_set_powergating_by_smu(struct amdgpu_device *adev, uint32_t block_type, bool gate)
{
int ret = 0;
bool swsmu = is_support_sw_smu(adev);
switch (block_type) {
case AMD_IP_BLOCK_TYPE_GFX:
case AMD_IP_BLOCK_TYPE_UVD:
case AMD_IP_BLOCK_TYPE_VCN:
case AMD_IP_BLOCK_TYPE_VCE:
if (swsmu)
ret = smu_dpm_set_power_gate(&adev->smu, block_type, gate);
else
ret = ((adev)->powerplay.pp_funcs->set_powergating_by_smu(
(adev)->powerplay.pp_handle, block_type, gate));
break;
case AMD_IP_BLOCK_TYPE_GMC:
case AMD_IP_BLOCK_TYPE_ACP:
case AMD_IP_BLOCK_TYPE_SDMA:
ret = ((adev)->powerplay.pp_funcs->set_powergating_by_smu(
(adev)->powerplay.pp_handle, block_type, gate));
break;
default:
break;
}
return ret;
}

7
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
Просмотреть файл

@ -355,10 +355,6 @@ enum amdgpu_pcie_gen {
((adev)->powerplay.pp_funcs->set_clockgating_by_smu(\
(adev)->powerplay.pp_handle, msg_id))
#define amdgpu_dpm_set_powergating_by_smu(adev, block_type, gate) \
((adev)->powerplay.pp_funcs->set_powergating_by_smu(\
(adev)->powerplay.pp_handle, block_type, gate))
#define amdgpu_dpm_get_power_profile_mode(adev, buf) \
((adev)->powerplay.pp_funcs->get_power_profile_mode(\
(adev)->powerplay.pp_handle, buf))
@ -520,6 +516,9 @@ enum amdgpu_pcie_gen amdgpu_get_pcie_gen_support(struct amdgpu_device *adev,
struct amd_vce_state*
amdgpu_get_vce_clock_state(void *handle, u32 idx);
int amdgpu_dpm_set_powergating_by_smu(struct amdgpu_device *adev,
uint32_t block_type, bool gate);
extern int amdgpu_dpm_get_sclk(struct amdgpu_device *adev, bool low);
extern int amdgpu_dpm_get_mclk(struct amdgpu_device *adev, bool low);

71
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
Просмотреть файл

@ -78,9 +78,10 @@
* - 3.30.0 - Add AMDGPU_SCHED_OP_CONTEXT_PRIORITY_OVERRIDE.
* - 3.31.0 - Add support for per-flip tiling attribute changes with DC
* - 3.32.0 - Add syncobj timeline support to AMDGPU_CS.
* - 3.33.0 - Fixes for GDS ENOMEM failures in AMDGPU_CS.
*/
#define KMS_DRIVER_MAJOR 3
#define KMS_DRIVER_MINOR 32
#define KMS_DRIVER_MINOR 33
#define KMS_DRIVER_PATCHLEVEL 0
#define AMDGPU_MAX_TIMEOUT_PARAM_LENTH 256
@ -110,7 +111,6 @@ int amdgpu_vm_fragment_size = -1;
int amdgpu_vm_block_size = -1;
int amdgpu_vm_fault_stop = 0;
int amdgpu_vm_debug = 0;
int amdgpu_vram_page_split = 512;
int amdgpu_vm_update_mode = -1;
int amdgpu_exp_hw_support = 0;
int amdgpu_dc = -1;
@ -138,6 +138,10 @@ int amdgpu_emu_mode = 0;
uint amdgpu_smu_memory_pool_size = 0;
/* FBC (bit 0) disabled by default*/
uint amdgpu_dc_feature_mask = 0;
int amdgpu_async_gfx_ring = 1;
int amdgpu_mcbp = 0;
int amdgpu_discovery = 0;
int amdgpu_mes = 0;
struct amdgpu_mgpu_info mgpu_info = {
.mutex = __MUTEX_INITIALIZER(mgpu_info.mutex),
@ -249,7 +253,9 @@ module_param_string(lockup_timeout, amdgpu_lockup_timeout, sizeof(amdgpu_lockup_
/**
* DOC: dpm (int)
* Override for dynamic power management setting (1 = enable, 0 = disable). The default is -1 (auto).
* Override for dynamic power management setting
* (0 = disable, 1 = enable, 2 = enable sw smu driver for vega20)
* The default is -1 (auto).
*/
MODULE_PARM_DESC(dpm, "DPM support (1 = enable, 0 = disable, -1 = auto)");
module_param_named(dpm, amdgpu_dpm, int, 0444);
@ -344,13 +350,6 @@ module_param_named(vm_debug, amdgpu_vm_debug, int, 0644);
MODULE_PARM_DESC(vm_update_mode, "VM update using CPU (0 = never (default except for large BAR(LB)), 1 = Graphics only, 2 = Compute only (default for LB), 3 = Both");
module_param_named(vm_update_mode, amdgpu_vm_update_mode, int, 0444);
/**
* DOC: vram_page_split (int)
* Override the number of pages after we split VRAM allocations (default 512, -1 = disable). The default is 512.
*/
MODULE_PARM_DESC(vram_page_split, "Number of pages after we split VRAM allocations (default 512, -1 = disable)");
module_param_named(vram_page_split, amdgpu_vram_page_split, int, 0444);
/**
* DOC: exp_hw_support (int)
* Enable experimental hw support (1 = enable). The default is 0 (disabled).
@ -574,6 +573,26 @@ MODULE_PARM_DESC(smu_memory_pool_size,
"0x1 = 256Mbyte, 0x2 = 512Mbyte, 0x4 = 1 Gbyte, 0x8 = 2GByte");
module_param_named(smu_memory_pool_size, amdgpu_smu_memory_pool_size, uint, 0444);
/**
* DOC: async_gfx_ring (int)
* It is used to enable gfx rings that could be configured with different prioritites or equal priorities
*/
MODULE_PARM_DESC(async_gfx_ring,
"Asynchronous GFX rings that could be configured with either different priorities (HP3D ring and LP3D ring), or equal priorities (0 = disabled, 1 = enabled (default))");
module_param_named(async_gfx_ring, amdgpu_async_gfx_ring, int, 0444);
MODULE_PARM_DESC(mcbp,
"Enable Mid-command buffer preemption (0 = disabled (default), 1 = enabled)");
module_param_named(mcbp, amdgpu_mcbp, int, 0444);
MODULE_PARM_DESC(discovery,
"Allow driver to discover hardware IPs from IP Discovery table at the top of VRAM");
module_param_named(discovery, amdgpu_discovery, int, 0444);
MODULE_PARM_DESC(mes,
"Enable Micro Engine Scheduler (0 = disabled (default), 1 = enabled)");
module_param_named(mes, amdgpu_mes, int, 0444);
#ifdef CONFIG_HSA_AMD
/**
* DOC: sched_policy (int)
@ -678,6 +697,14 @@ MODULE_PARM_DESC(halt_if_hws_hang, "Halt if HWS hang is detected (0 = off (defau
bool hws_gws_support;
module_param(hws_gws_support, bool, 0444);
MODULE_PARM_DESC(hws_gws_support, "MEC FW support gws barriers (false = not supported (Default), true = supported)");
/**
* DOC: queue_preemption_timeout_ms (int)
* queue preemption timeout in ms (1 = Minimum, 9000 = default)
*/
int queue_preemption_timeout_ms;
module_param(queue_preemption_timeout_ms, int, 0644);
MODULE_PARM_DESC(queue_preemption_timeout_ms, "queue preemption timeout in ms (1 = Minimum, 9000 = default)");
#endif
/**
@ -688,6 +715,22 @@ MODULE_PARM_DESC(hws_gws_support, "MEC FW support gws barriers (false = not supp
MODULE_PARM_DESC(dcfeaturemask, "all stable DC features enabled (default))");
module_param_named(dcfeaturemask, amdgpu_dc_feature_mask, uint, 0444);
/**
* DOC: abmlevel (uint)
* Override the default ABM (Adaptive Backlight Management) level used for DC
* enabled hardware. Requires DMCU to be supported and loaded.
* Valid levels are 0-4. A value of 0 indicates that ABM should be disabled by
* default. Values 1-4 control the maximum allowable brightness reduction via
* the ABM algorithm, with 1 being the least reduction and 4 being the most
* reduction.
*
* Defaults to 0, or disabled. Userspace can still override this level later
* after boot.
*/
uint amdgpu_dm_abm_level = 0;
MODULE_PARM_DESC(abmlevel, "ABM level (0 = off (default), 1-4 = backlight reduction level) ");
module_param_named(abmlevel, amdgpu_dm_abm_level, uint, 0444);
static const struct pci_device_id pciidlist[] = {
#ifdef CONFIG_DRM_AMDGPU_SI
{0x1002, 0x6780, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_TAHITI},
@ -944,6 +987,14 @@ static const struct pci_device_id pciidlist[] = {
/* Raven */
{0x1002, 0x15dd, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU},
{0x1002, 0x15d8, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RAVEN|AMD_IS_APU},
/* Navi10 */
{0x1002, 0x7310, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0x1002, 0x7312, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0x1002, 0x7318, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0x1002, 0x7319, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0x1002, 0x731A, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0x1002, 0x731B, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0x1002, 0x731F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI10},
{0, 0, 0}
};

18
drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
Просмотреть файл

@ -709,22 +709,30 @@ static int amdgpu_debugfs_fence_info(struct seq_file *m, void *data)
amdgpu_fence_process(ring);
seq_printf(m, "--- ring %d (%s) ---\n", i, ring->name);
seq_printf(m, "Last signaled fence 0x%08x\n",
seq_printf(m, "Last signaled fence 0x%08x\n",
atomic_read(&ring->fence_drv.last_seq));
seq_printf(m, "Last emitted 0x%08x\n",
seq_printf(m, "Last emitted 0x%08x\n",
ring->fence_drv.sync_seq);
if (ring->funcs->type == AMDGPU_RING_TYPE_GFX ||
ring->funcs->type == AMDGPU_RING_TYPE_SDMA) {
seq_printf(m, "Last signaled trailing fence 0x%08x\n",
le32_to_cpu(*ring->trail_fence_cpu_addr));
seq_printf(m, "Last emitted 0x%08x\n",
ring->trail_seq);
}
if (ring->funcs->type != AMDGPU_RING_TYPE_GFX)
continue;
/* set in CP_VMID_PREEMPT and preemption occurred */
seq_printf(m, "Last preempted 0x%08x\n",
seq_printf(m, "Last preempted 0x%08x\n",
le32_to_cpu(*(ring->fence_drv.cpu_addr + 2)));
/* set in CP_VMID_RESET and reset occurred */
seq_printf(m, "Last reset 0x%08x\n",
seq_printf(m, "Last reset 0x%08x\n",
le32_to_cpu(*(ring->fence_drv.cpu_addr + 4)));
/* Both preemption and reset occurred */
seq_printf(m, "Last both 0x%08x\n",
seq_printf(m, "Last both 0x%08x\n",
le32_to_cpu(*(ring->fence_drv.cpu_addr + 6)));
}
return 0;

4
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
Просмотреть файл

@ -175,7 +175,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, true);
r = ttm_eu_reserve_buffers(&ticket, &list, false, &duplicates, false);
if (r) {
dev_err(adev->dev, "leaking bo va because "
"we fail to reserve bo (%d)\n", r);
@ -612,7 +612,7 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
amdgpu_vm_get_pd_bo(&fpriv->vm, &list, &vm_pd);
r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, true);
r = ttm_eu_reserve_buffers(&ticket, &list, true, &duplicates, false);
if (r)
goto error_unref;

180
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
Просмотреть файл

@ -34,8 +34,8 @@
* GPU GFX IP block helpers function.
*/
int amdgpu_gfx_queue_to_bit(struct amdgpu_device *adev, int mec,
int pipe, int queue)
int amdgpu_gfx_mec_queue_to_bit(struct amdgpu_device *adev, int mec,
int pipe, int queue)
{
int bit = 0;
@ -47,8 +47,8 @@ int amdgpu_gfx_queue_to_bit(struct amdgpu_device *adev, int mec,
return bit;
}
void amdgpu_gfx_bit_to_queue(struct amdgpu_device *adev, int bit,
int *mec, int *pipe, int *queue)
void amdgpu_gfx_bit_to_mec_queue(struct amdgpu_device *adev, int bit,
int *mec, int *pipe, int *queue)
{
*queue = bit % adev->gfx.mec.num_queue_per_pipe;
*pipe = (bit / adev->gfx.mec.num_queue_per_pipe)
@ -61,10 +61,40 @@ void amdgpu_gfx_bit_to_queue(struct amdgpu_device *adev, int bit,
bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev,
int mec, int pipe, int queue)
{
return test_bit(amdgpu_gfx_queue_to_bit(adev, mec, pipe, queue),
return test_bit(amdgpu_gfx_mec_queue_to_bit(adev, mec, pipe, queue),
adev->gfx.mec.queue_bitmap);
}
int amdgpu_gfx_me_queue_to_bit(struct amdgpu_device *adev,
int me, int pipe, int queue)
{
int bit = 0;
bit += me * adev->gfx.me.num_pipe_per_me
* adev->gfx.me.num_queue_per_pipe;
bit += pipe * adev->gfx.me.num_queue_per_pipe;
bit += queue;
return bit;
}
void amdgpu_gfx_bit_to_me_queue(struct amdgpu_device *adev, int bit,
int *me, int *pipe, int *queue)
{
*queue = bit % adev->gfx.me.num_queue_per_pipe;
*pipe = (bit / adev->gfx.me.num_queue_per_pipe)
% adev->gfx.me.num_pipe_per_me;
*me = (bit / adev->gfx.me.num_queue_per_pipe)
/ adev->gfx.me.num_pipe_per_me;
}
bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev,
int me, int pipe, int queue)
{
return test_bit(amdgpu_gfx_me_queue_to_bit(adev, me, pipe, queue),
adev->gfx.me.queue_bitmap);
}
/**
* amdgpu_gfx_scratch_get - Allocate a scratch register
*
@ -199,6 +229,30 @@ void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev)
adev->gfx.num_compute_rings = AMDGPU_MAX_COMPUTE_RINGS;
}
void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev)
{
int i, queue, pipe, me;
for (i = 0; i < AMDGPU_MAX_GFX_QUEUES; ++i) {
queue = i % adev->gfx.me.num_queue_per_pipe;
pipe = (i / adev->gfx.me.num_queue_per_pipe)
% adev->gfx.me.num_pipe_per_me;
me = (i / adev->gfx.me.num_queue_per_pipe)
/ adev->gfx.me.num_pipe_per_me;
if (me >= adev->gfx.me.num_me)
break;
/* policy: amdgpu owns the first queue per pipe at this stage
* will extend to mulitple queues per pipe later */
if (me == 0 && queue < 1)
set_bit(i, adev->gfx.me.queue_bitmap);
}
/* update the number of active graphics rings */
adev->gfx.num_gfx_rings =
bitmap_weight(adev->gfx.me.queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
}
static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
struct amdgpu_ring *ring)
{
@ -213,7 +267,7 @@ static int amdgpu_gfx_kiq_acquire(struct amdgpu_device *adev,
if (test_bit(queue_bit, adev->gfx.mec.queue_bitmap))
continue;
amdgpu_gfx_bit_to_queue(adev, queue_bit, &mec, &pipe, &queue);
amdgpu_gfx_bit_to_mec_queue(adev, queue_bit, &mec, &pipe, &queue);
/*
* 1. Using pipes 2/3 from MEC 2 seems cause problems.
@ -306,9 +360,9 @@ int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
return 0;
}
/* create MQD for each compute queue */
int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
unsigned mqd_size)
/* create MQD for each compute/gfx queue */
int amdgpu_gfx_mqd_sw_init(struct amdgpu_device *adev,
unsigned mqd_size)
{
struct amdgpu_ring *ring = NULL;
int r, i;
@ -335,6 +389,27 @@ int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
}
if (adev->asic_type == CHIP_NAVI10 && amdgpu_async_gfx_ring) {
/* create MQD for each KGQ */
for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
ring = &adev->gfx.gfx_ring[i];
if (!ring->mqd_obj) {
r = amdgpu_bo_create_kernel(adev, mqd_size, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
&ring->mqd_gpu_addr, &ring->mqd_ptr);
if (r) {
dev_warn(adev->dev, "failed to create ring mqd bo (%d)", r);
return r;
}
/* prepare MQD backup */
adev->gfx.me.mqd_backup[i] = kmalloc(mqd_size, GFP_KERNEL);
if (!adev->gfx.me.mqd_backup[i])
dev_warn(adev->dev, "no memory to create MQD backup for ring %s\n", ring->name);
}
}
}
/* create MQD for each KCQ */
for (i = 0; i < adev->gfx.num_compute_rings; i++) {
ring = &adev->gfx.compute_ring[i];
@ -343,7 +418,7 @@ int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
&ring->mqd_gpu_addr, &ring->mqd_ptr);
if (r) {
dev_warn(adev->dev, "failed to create ring mqd ob (%d)", r);
dev_warn(adev->dev, "failed to create ring mqd bo (%d)", r);
return r;
}
@ -357,11 +432,21 @@ int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
return 0;
}
void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev)
void amdgpu_gfx_mqd_sw_fini(struct amdgpu_device *adev)
{
struct amdgpu_ring *ring = NULL;
int i;
if (adev->asic_type == CHIP_NAVI10 && amdgpu_async_gfx_ring) {
for (i = 0; i < adev->gfx.num_gfx_rings; i++) {
ring = &adev->gfx.gfx_ring[i];
kfree(adev->gfx.me.mqd_backup[i]);
amdgpu_bo_free_kernel(&ring->mqd_obj,
&ring->mqd_gpu_addr,
&ring->mqd_ptr);
}
}
for (i = 0; i < adev->gfx.num_compute_rings; i++) {
ring = &adev->gfx.compute_ring[i];
kfree(adev->gfx.mec.mqd_backup[i]);
@ -371,12 +456,81 @@ void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev)
}
ring = &adev->gfx.kiq.ring;
if (adev->asic_type == CHIP_NAVI10 && amdgpu_async_gfx_ring)
kfree(adev->gfx.me.mqd_backup[AMDGPU_MAX_GFX_RINGS]);
kfree(adev->gfx.mec.mqd_backup[AMDGPU_MAX_COMPUTE_RINGS]);
amdgpu_bo_free_kernel(&ring->mqd_obj,
&ring->mqd_gpu_addr,
&ring->mqd_ptr);
}
int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev)
{
struct amdgpu_kiq *kiq = &adev->gfx.kiq;
struct amdgpu_ring *kiq_ring = &kiq->ring;
int i;
if (!kiq->pmf || !kiq->pmf->kiq_unmap_queues)
return -EINVAL;
if (amdgpu_ring_alloc(kiq_ring, kiq->pmf->unmap_queues_size *
adev->gfx.num_compute_rings))
return -ENOMEM;
for (i = 0; i < adev->gfx.num_compute_rings; i++)
kiq->pmf->kiq_unmap_queues(kiq_ring, &adev->gfx.compute_ring[i],
RESET_QUEUES, 0, 0);
return amdgpu_ring_test_ring(kiq_ring);
}
int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev)
{
struct amdgpu_kiq *kiq = &adev->gfx.kiq;
struct amdgpu_ring *kiq_ring = &adev->gfx.kiq.ring;
uint64_t queue_mask = 0;
int r, i;
if (!kiq->pmf || !kiq->pmf->kiq_map_queues || !kiq->pmf->kiq_set_resources)
return -EINVAL;
for (i = 0; i < AMDGPU_MAX_COMPUTE_QUEUES; ++i) {
if (!test_bit(i, adev->gfx.mec.queue_bitmap))
continue;
/* This situation may be hit in the future if a new HW
* generation exposes more than 64 queues. If so, the
* definition of queue_mask needs updating */
if (WARN_ON(i > (sizeof(queue_mask)*8))) {
DRM_ERROR("Invalid KCQ enabled: %d\n", i);
break;
}
queue_mask |= (1ull << i);
}
DRM_INFO("kiq ring mec %d pipe %d q %d\n", kiq_ring->me, kiq_ring->pipe,
kiq_ring->queue);
r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size *
adev->gfx.num_compute_rings +
kiq->pmf->set_resources_size);
if (r) {
DRM_ERROR("Failed to lock KIQ (%d).\n", r);
return r;
}
kiq->pmf->kiq_set_resources(kiq_ring, queue_mask);
for (i = 0; i < adev->gfx.num_compute_rings; i++)
kiq->pmf->kiq_map_queues(kiq_ring, &adev->gfx.compute_ring[i]);
r = amdgpu_ring_test_helper(kiq_ring);
if (r)
DRM_ERROR("KCQ enable failed\n");
return r;
}
/* amdgpu_gfx_off_ctrl - Handle gfx off feature enable/disable
*
* @adev: amdgpu_device pointer
@ -393,7 +547,9 @@ void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable)
if (!(adev->pm.pp_feature & PP_GFXOFF_MASK))
return;
if (!adev->powerplay.pp_funcs || !adev->powerplay.pp_funcs->set_powergating_by_smu)
if (!is_support_sw_smu(adev) &&
(!adev->powerplay.pp_funcs ||
!adev->powerplay.pp_funcs->set_powergating_by_smu))
return;

86
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
Просмотреть файл

@ -38,6 +38,7 @@
#define AMDGPU_GFX_CG_DISABLED_MODE 0x00000004L
#define AMDGPU_GFX_LBPW_DISABLED_MODE 0x00000008L
#define AMDGPU_MAX_GFX_QUEUES KGD_MAX_QUEUES
#define AMDGPU_MAX_COMPUTE_QUEUES KGD_MAX_QUEUES
struct amdgpu_mec {
@ -54,12 +55,41 @@ struct amdgpu_mec {
DECLARE_BITMAP(queue_bitmap, AMDGPU_MAX_COMPUTE_QUEUES);
};
enum amdgpu_unmap_queues_action {
PREEMPT_QUEUES = 0,
RESET_QUEUES,
DISABLE_PROCESS_QUEUES,
PREEMPT_QUEUES_NO_UNMAP,
};
struct kiq_pm4_funcs {
/* Support ASIC-specific kiq pm4 packets*/
void (*kiq_set_resources)(struct amdgpu_ring *kiq_ring,
uint64_t queue_mask);
void (*kiq_map_queues)(struct amdgpu_ring *kiq_ring,
struct amdgpu_ring *ring);
void (*kiq_unmap_queues)(struct amdgpu_ring *kiq_ring,
struct amdgpu_ring *ring,
enum amdgpu_unmap_queues_action action,
u64 gpu_addr, u64 seq);
void (*kiq_query_status)(struct amdgpu_ring *kiq_ring,
struct amdgpu_ring *ring,
u64 addr,
u64 seq);
/* Packet sizes */
int set_resources_size;
int map_queues_size;
int unmap_queues_size;
int query_status_size;
};
struct amdgpu_kiq {
u64 eop_gpu_addr;
struct amdgpu_bo *eop_obj;
spinlock_t ring_lock;
struct amdgpu_ring ring;
struct amdgpu_irq_src irq;
const struct kiq_pm4_funcs *pmf;
};
/*
@ -131,6 +161,10 @@ struct amdgpu_gfx_config {
uint32_t double_offchip_lds_buf;
/* cached value of DB_DEBUG2 */
uint32_t db_debug2;
/* gfx10 specific config */
uint32_t num_sc_per_sh;
uint32_t num_packer_per_sc;
uint32_t pa_sc_tile_steering_override;
};
struct amdgpu_cu_info {
@ -191,10 +225,38 @@ struct sq_work {
unsigned ih_data;
};
struct amdgpu_pfp {
struct amdgpu_bo *pfp_fw_obj;
uint64_t pfp_fw_gpu_addr;
uint32_t *pfp_fw_ptr;
};
struct amdgpu_ce {
struct amdgpu_bo *ce_fw_obj;
uint64_t ce_fw_gpu_addr;
uint32_t *ce_fw_ptr;
};
struct amdgpu_me {
struct amdgpu_bo *me_fw_obj;
uint64_t me_fw_gpu_addr;
uint32_t *me_fw_ptr;
uint32_t num_me;
uint32_t num_pipe_per_me;
uint32_t num_queue_per_pipe;
void *mqd_backup[AMDGPU_MAX_GFX_RINGS + 1];
/* These are the resources for which amdgpu takes ownership */
DECLARE_BITMAP(queue_bitmap, AMDGPU_MAX_GFX_QUEUES);
};
struct amdgpu_gfx {
struct mutex gpu_clock_mutex;
struct amdgpu_gfx_config config;
struct amdgpu_rlc rlc;
struct amdgpu_pfp pfp;
struct amdgpu_ce ce;
struct amdgpu_me me;
struct amdgpu_mec mec;
struct amdgpu_kiq kiq;
struct amdgpu_scratch scratch;
@ -297,17 +359,27 @@ void amdgpu_gfx_kiq_fini(struct amdgpu_device *adev);
int amdgpu_gfx_kiq_init(struct amdgpu_device *adev,
unsigned hpd_size);
int amdgpu_gfx_compute_mqd_sw_init(struct amdgpu_device *adev,
unsigned mqd_size);
void amdgpu_gfx_compute_mqd_sw_fini(struct amdgpu_device *adev);
int amdgpu_gfx_mqd_sw_init(struct amdgpu_device *adev,
unsigned mqd_size);
void amdgpu_gfx_mqd_sw_fini(struct amdgpu_device *adev);
int amdgpu_gfx_disable_kcq(struct amdgpu_device *adev);
int amdgpu_gfx_enable_kcq(struct amdgpu_device *adev);
void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev);
int amdgpu_gfx_queue_to_bit(struct amdgpu_device *adev, int mec,
int pipe, int queue);
void amdgpu_gfx_bit_to_queue(struct amdgpu_device *adev, int bit,
int *mec, int *pipe, int *queue);
void amdgpu_gfx_graphics_queue_acquire(struct amdgpu_device *adev);
int amdgpu_gfx_mec_queue_to_bit(struct amdgpu_device *adev, int mec,
int pipe, int queue);
void amdgpu_gfx_bit_to_mec_queue(struct amdgpu_device *adev, int bit,
int *mec, int *pipe, int *queue);
bool amdgpu_gfx_is_mec_queue_enabled(struct amdgpu_device *adev, int mec,
int pipe, int queue);
int amdgpu_gfx_me_queue_to_bit(struct amdgpu_device *adev, int me,
int pipe, int queue);
void amdgpu_gfx_bit_to_me_queue(struct amdgpu_device *adev, int bit,
int *me, int *pipe, int *queue);
bool amdgpu_gfx_is_me_queue_enabled(struct amdgpu_device *adev, int me,
int pipe, int queue);
void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable);
#endif

8
drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
Просмотреть файл

@ -211,6 +211,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
skip_preamble = ring->current_ctx == fence_ctx;
if (job && ring->funcs->emit_cntxcntl) {
status |= job->preamble_status;
status |= job->preemption_status;
amdgpu_ring_emit_cntxcntl(ring, status);
}
@ -219,9 +220,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
/* drop preamble IBs if we don't have a context switch */
if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) &&
skip_preamble &&
!(status & AMDGPU_PREAMBLE_IB_PRESENT_FIRST) &&
!amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */
skip_preamble &&
!(status & AMDGPU_PREAMBLE_IB_PRESENT_FIRST) &&
!amdgpu_mcbp &&
!amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */
continue;
amdgpu_ring_emit_ib(ring, job, ib, status);

7
drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c
Просмотреть файл

@ -364,8 +364,11 @@ static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm,
if (updates && (!flushed || dma_fence_is_later(updates, flushed)))
needs_flush = true;
/* Concurrent flushes are only possible starting with Vega10 */
if (adev->asic_type < CHIP_VEGA10 && needs_flush)
/* Concurrent flushes are only possible starting with Vega10 and
* are broken on Navi10 and Navi14.
*/
if (needs_flush && (adev->asic_type < CHIP_VEGA10 ||
adev->asic_type == CHIP_NAVI10))
continue;
/* Good, we can use this VMID. Remember this submission as

3
drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
Просмотреть файл

@ -29,6 +29,8 @@
#define AMDGPU_PREAMBLE_IB_PRESENT_FIRST (1 << 1)
/* bit set means context switch occured */
#define AMDGPU_HAVE_CTX_SWITCH (1 << 2)
/* bit set means IB is preempted */
#define AMDGPU_IB_PREEMPTED (1 << 3)
#define to_amdgpu_job(sched_job) \
container_of((sched_job), struct amdgpu_job, base)
@ -45,6 +47,7 @@ struct amdgpu_job {
struct amdgpu_ib *ibs;
struct dma_fence *fence; /* the hw fence */
uint32_t preamble_status;
uint32_t preemption_status;
uint32_t num_ibs;
void *owner;
bool vm_needs_flush;

10
drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
Просмотреть файл

@ -712,7 +712,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
dev_info.ids_flags = 0;
if (adev->flags & AMD_IS_APU)
dev_info.ids_flags |= AMDGPU_IDS_FLAGS_FUSION;
if (amdgpu_sriov_vf(adev))
if (amdgpu_mcbp || amdgpu_sriov_vf(adev))
dev_info.ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION;
vm_size = adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
@ -765,6 +765,10 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
dev_info.gs_prim_buffer_depth = adev->gfx.config.gs_prim_buffer_depth;
dev_info.max_gs_waves_per_vgt = adev->gfx.config.max_gs_threads;
if (adev->family >= AMDGPU_FAMILY_NV)
dev_info.pa_sc_tile_steering_override =
adev->gfx.config.pa_sc_tile_steering_override;
return copy_to_user(out, &dev_info,
min((size_t)size, sizeof(dev_info))) ? -EFAULT : 0;
}
@ -1006,7 +1010,7 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv)
goto error_vm;
}
if (amdgpu_sriov_vf(adev)) {
if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK;
r = amdgpu_map_static_csa(adev, &fpriv->vm, adev->virt.csa_obj,
@ -1069,7 +1073,7 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev,
amdgpu_vm_bo_rmv(adev, fpriv->prt_va);
if (amdgpu_sriov_vf(adev)) {
if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) {
/* TODO: how to handle reserve failure */
BUG_ON(amdgpu_bo_reserve(adev->virt.csa_obj, true));
amdgpu_vm_bo_rmv(adev, fpriv->csa_va);

101
drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h Normal file
Просмотреть файл

@ -0,0 +1,101 @@
/*
* Copyright 2019 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 __AMDGPU_MES_H__
#define __AMDGPU_MES_H__
struct amdgpu_mes_funcs;
struct amdgpu_mes {
struct amdgpu_adev *adev;
const struct firmware *fw;
/* mes ucode */
struct amdgpu_bo *ucode_fw_obj;
uint64_t ucode_fw_gpu_addr;
uint32_t *ucode_fw_ptr;
uint32_t ucode_fw_version;
uint64_t uc_start_addr;
/* mes ucode data */
struct amdgpu_bo *data_fw_obj;
uint64_t data_fw_gpu_addr;
uint32_t *data_fw_ptr;
uint32_t data_fw_version;
uint64_t data_start_addr;
/* ip specific functions */
struct amdgpu_mes_funcs *funcs;
};
struct mes_add_queue_input {
uint32_t process_id;
uint64_t page_table_base_addr;
uint64_t process_va_start;
uint64_t process_va_end;
uint64_t process_quantum;
uint64_t process_context_addr;
uint64_t gang_quantum;
uint64_t gang_context_addr;
uint32_t inprocess_gang_priority;
uint32_t gang_global_priority_level;
uint32_t doorbell_offset;
uint64_t mqd_addr;
uint64_t wptr_addr;
uint32_t queue_type;
uint32_t paging;
};
struct mes_remove_queue_input {
uint32_t doorbell_offset;
uint64_t gang_context_addr;
};
struct mes_suspend_gang_input {
bool suspend_all_gangs;
uint64_t gang_context_addr;
uint64_t suspend_fence_addr;
uint32_t suspend_fence_value;
};
struct mes_resume_gang_input {
bool resume_all_gangs;
uint64_t gang_context_addr;
};
struct amdgpu_mes_funcs {
int (*add_hw_queue)(struct amdgpu_mes *mes,
struct mes_add_queue_input *input);
int (*remove_hw_queue)(struct amdgpu_mes *mes,
struct mes_remove_queue_input *input);
int (*suspend_gang)(struct amdgpu_mes *mes,
struct mes_suspend_gang_input *input);
int (*resume_gang)(struct amdgpu_mes *mes,
struct mes_resume_gang_input *input);
};
#endif /* __AMDGPU_MES_H__ */

1
drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
Просмотреть файл

@ -519,7 +519,6 @@ void amdgpu_hmm_init_range(struct hmm_range *range)
range->flags = hmm_range_flags;
range->values = hmm_range_values;
range->pfn_shift = PAGE_SHIFT;
range->pfns = NULL;
INIT_LIST_HEAD(&range->list);
}
}

7
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
Просмотреть файл

@ -495,7 +495,11 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev,
#endif
bo->tbo.bdev = &adev->mman.bdev;
amdgpu_bo_placement_from_domain(bo, bp->domain);
if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA |
AMDGPU_GEM_DOMAIN_GDS))
amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU);
else
amdgpu_bo_placement_from_domain(bo, bp->domain);
if (bp->type == ttm_bo_type_kernel)
bo->tbo.priority = 1;
@ -975,6 +979,7 @@ static const char *amdgpu_vram_names[] = {
"HBM",
"DDR3",
"DDR4",
"GDDR6",
};
/**

2
drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
Просмотреть файл

@ -155,7 +155,7 @@ static inline int amdgpu_bo_reserve(struct amdgpu_bo *bo, bool no_intr)
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
int r;
r = ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL);
r = __ttm_bo_reserve(&bo->tbo, !no_intr, false, NULL);
if (unlikely(r != 0)) {
if (r != -ERESTARTSYS)
dev_err(adev->dev, "%p reserve failed\n", bo);

127
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
Просмотреть файл

@ -67,6 +67,9 @@ static const struct cg_flag_name clocks[] = {
{AMD_CG_SUPPORT_DRM_LS, "Digital Right Management Light Sleep"},
{AMD_CG_SUPPORT_ROM_MGCG, "Rom Medium Grain Clock Gating"},
{AMD_CG_SUPPORT_DF_MGCG, "Data Fabric Medium Grain Clock Gating"},
{AMD_CG_SUPPORT_ATHUB_MGCG, "Address Translation Hub Medium Grain Clock Gating"},
{AMD_CG_SUPPORT_ATHUB_LS, "Address Translation Hub Light Sleep"},
{0, NULL},
};
@ -272,8 +275,11 @@ static ssize_t amdgpu_get_dpm_forced_performance_level(struct device *dev,
struct amdgpu_device *adev = ddev->dev_private;
enum amd_dpm_forced_level level = 0xff;
if ((adev->flags & AMD_IS_PX) &&
(ddev->switch_power_state != DRM_SWITCH_POWER_ON))
if (amdgpu_sriov_vf(adev))
return 0;
if ((adev->flags & AMD_IS_PX) &&
(ddev->switch_power_state != DRM_SWITCH_POWER_ON))
return snprintf(buf, PAGE_SIZE, "off\n");
if (is_support_sw_smu(adev))
@ -311,10 +317,12 @@ static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev,
(ddev->switch_power_state != DRM_SWITCH_POWER_ON))
return -EINVAL;
if (is_support_sw_smu(adev))
current_level = smu_get_performance_level(&adev->smu);
else if (adev->powerplay.pp_funcs->get_performance_level)
current_level = amdgpu_dpm_get_performance_level(adev);
if (!amdgpu_sriov_vf(adev)) {
if (is_support_sw_smu(adev))
current_level = smu_get_performance_level(&adev->smu);
else if (adev->powerplay.pp_funcs->get_performance_level)
current_level = amdgpu_dpm_get_performance_level(adev);
}
if (strncmp("low", buf, strlen("low")) == 0) {
level = AMD_DPM_FORCED_LEVEL_LOW;
@ -365,18 +373,9 @@ static ssize_t amdgpu_set_dpm_forced_performance_level(struct device *dev,
}
if (is_support_sw_smu(adev)) {
mutex_lock(&adev->pm.mutex);
if (adev->pm.dpm.thermal_active) {
count = -EINVAL;
mutex_unlock(&adev->pm.mutex);
goto fail;
}
ret = smu_force_performance_level(&adev->smu, level);
if (ret)
count = -EINVAL;
else
adev->pm.dpm.forced_level = level;
mutex_unlock(&adev->pm.mutex);
} else if (adev->powerplay.pp_funcs->force_performance_level) {
mutex_lock(&adev->pm.mutex);
if (adev->pm.dpm.thermal_active) {
@ -690,12 +689,12 @@ static ssize_t amdgpu_set_pp_od_clk_voltage(struct device *dev,
if (ret)
return -EINVAL;
} else {
if (adev->powerplay.pp_funcs->odn_edit_dpm_table)
if (adev->powerplay.pp_funcs->odn_edit_dpm_table) {
ret = amdgpu_dpm_odn_edit_dpm_table(adev, type,
parameter, parameter_size);
if (ret)
return -EINVAL;
if (ret)
return -EINVAL;
}
if (type == PP_OD_COMMIT_DPM_TABLE) {
if (adev->powerplay.pp_funcs->dispatch_tasks) {
@ -721,10 +720,10 @@ static ssize_t amdgpu_get_pp_od_clk_voltage(struct device *dev,
uint32_t size = 0;
if (is_support_sw_smu(adev)) {
size = smu_print_clk_levels(&adev->smu, OD_SCLK, buf);
size += smu_print_clk_levels(&adev->smu, OD_MCLK, buf+size);
size += smu_print_clk_levels(&adev->smu, OD_VDDC_CURVE, buf+size);
size += smu_print_clk_levels(&adev->smu, OD_RANGE, buf+size);
size = smu_print_clk_levels(&adev->smu, SMU_OD_SCLK, buf);
size += smu_print_clk_levels(&adev->smu, SMU_OD_MCLK, buf+size);
size += smu_print_clk_levels(&adev->smu, SMU_OD_VDDC_CURVE, buf+size);
size += smu_print_clk_levels(&adev->smu, SMU_OD_RANGE, buf+size);
return size;
} else if (adev->powerplay.pp_funcs->print_clock_levels) {
size = amdgpu_dpm_print_clock_levels(adev, OD_SCLK, buf);
@ -835,7 +834,7 @@ static ssize_t amdgpu_get_pp_dpm_sclk(struct device *dev,
return adev->virt.ops->get_pp_clk(adev, PP_SCLK, buf);
if (is_support_sw_smu(adev))
return smu_print_clk_levels(&adev->smu, PP_SCLK, buf);
return smu_print_clk_levels(&adev->smu, SMU_SCLK, buf);
else if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_SCLK, buf);
else
@ -888,12 +887,15 @@ static ssize_t amdgpu_set_pp_dpm_sclk(struct device *dev,
int ret;
uint32_t mask = 0;
if (amdgpu_sriov_vf(adev))
return 0;
ret = amdgpu_read_mask(buf, count, &mask);
if (ret)
return ret;
if (is_support_sw_smu(adev))
ret = smu_force_clk_levels(&adev->smu, PP_SCLK, mask);
ret = smu_force_clk_levels(&adev->smu, SMU_SCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_SCLK, mask);
@ -910,8 +912,12 @@ static ssize_t amdgpu_get_pp_dpm_mclk(struct device *dev,
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
if (amdgpu_sriov_vf(adev) && amdgim_is_hwperf(adev) &&
adev->virt.ops->get_pp_clk)
return adev->virt.ops->get_pp_clk(adev, PP_MCLK, buf);
if (is_support_sw_smu(adev))
return smu_print_clk_levels(&adev->smu, PP_MCLK, buf);
return smu_print_clk_levels(&adev->smu, SMU_MCLK, buf);
else if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_MCLK, buf);
else
@ -928,12 +934,15 @@ static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
int ret;
uint32_t mask = 0;
if (amdgpu_sriov_vf(adev))
return 0;
ret = amdgpu_read_mask(buf, count, &mask);
if (ret)
return ret;
if (is_support_sw_smu(adev))
ret = smu_force_clk_levels(&adev->smu, PP_MCLK, mask);
ret = smu_force_clk_levels(&adev->smu, SMU_MCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_MCLK, mask);
@ -951,7 +960,7 @@ static ssize_t amdgpu_get_pp_dpm_socclk(struct device *dev,
struct amdgpu_device *adev = ddev->dev_private;
if (is_support_sw_smu(adev))
return smu_print_clk_levels(&adev->smu, PP_SOCCLK, buf);
return smu_print_clk_levels(&adev->smu, SMU_SOCCLK, buf);
else if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_SOCCLK, buf);
else
@ -973,7 +982,7 @@ static ssize_t amdgpu_set_pp_dpm_socclk(struct device *dev,
return ret;
if (is_support_sw_smu(adev))
ret = smu_force_clk_levels(&adev->smu, PP_SOCCLK, mask);
ret = smu_force_clk_levels(&adev->smu, SMU_SOCCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_SOCCLK, mask);
@ -991,7 +1000,7 @@ static ssize_t amdgpu_get_pp_dpm_fclk(struct device *dev,
struct amdgpu_device *adev = ddev->dev_private;
if (is_support_sw_smu(adev))
return smu_print_clk_levels(&adev->smu, PP_FCLK, buf);
return smu_print_clk_levels(&adev->smu, SMU_FCLK, buf);
else if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_FCLK, buf);
else
@ -1013,7 +1022,7 @@ static ssize_t amdgpu_set_pp_dpm_fclk(struct device *dev,
return ret;
if (is_support_sw_smu(adev))
ret = smu_force_clk_levels(&adev->smu, PP_FCLK, mask);
ret = smu_force_clk_levels(&adev->smu, SMU_FCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_FCLK, mask);
@ -1031,7 +1040,7 @@ static ssize_t amdgpu_get_pp_dpm_dcefclk(struct device *dev,
struct amdgpu_device *adev = ddev->dev_private;
if (is_support_sw_smu(adev))
return smu_print_clk_levels(&adev->smu, PP_DCEFCLK, buf);
return smu_print_clk_levels(&adev->smu, SMU_DCEFCLK, buf);
else if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_DCEFCLK, buf);
else
@ -1053,7 +1062,7 @@ static ssize_t amdgpu_set_pp_dpm_dcefclk(struct device *dev,
return ret;
if (is_support_sw_smu(adev))
ret = smu_force_clk_levels(&adev->smu, PP_DCEFCLK, mask);
ret = smu_force_clk_levels(&adev->smu, SMU_DCEFCLK, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_DCEFCLK, mask);
@ -1071,7 +1080,7 @@ static ssize_t amdgpu_get_pp_dpm_pcie(struct device *dev,
struct amdgpu_device *adev = ddev->dev_private;
if (is_support_sw_smu(adev))
return smu_print_clk_levels(&adev->smu, PP_PCIE, buf);
return smu_print_clk_levels(&adev->smu, SMU_PCIE, buf);
else if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_PCIE, buf);
else
@ -1093,7 +1102,7 @@ static ssize_t amdgpu_set_pp_dpm_pcie(struct device *dev,
return ret;
if (is_support_sw_smu(adev))
ret = smu_force_clk_levels(&adev->smu, PP_PCIE, mask);
ret = smu_force_clk_levels(&adev->smu, SMU_PCIE, mask);
else if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_PCIE, mask);
@ -1112,7 +1121,7 @@ static ssize_t amdgpu_get_pp_sclk_od(struct device *dev,
uint32_t value = 0;
if (is_support_sw_smu(adev))
value = smu_get_od_percentage(&(adev->smu), OD_SCLK);
value = smu_get_od_percentage(&(adev->smu), SMU_OD_SCLK);
else if (adev->powerplay.pp_funcs->get_sclk_od)
value = amdgpu_dpm_get_sclk_od(adev);
@ -1137,7 +1146,7 @@ static ssize_t amdgpu_set_pp_sclk_od(struct device *dev,
}
if (is_support_sw_smu(adev)) {
value = smu_set_od_percentage(&(adev->smu), OD_SCLK, (uint32_t)value);
value = smu_set_od_percentage(&(adev->smu), SMU_OD_SCLK, (uint32_t)value);
} else {
if (adev->powerplay.pp_funcs->set_sclk_od)
amdgpu_dpm_set_sclk_od(adev, (uint32_t)value);
@ -1163,7 +1172,7 @@ static ssize_t amdgpu_get_pp_mclk_od(struct device *dev,
uint32_t value = 0;
if (is_support_sw_smu(adev))
value = smu_get_od_percentage(&(adev->smu), OD_MCLK);
value = smu_get_od_percentage(&(adev->smu), SMU_OD_MCLK);
else if (adev->powerplay.pp_funcs->get_mclk_od)
value = amdgpu_dpm_get_mclk_od(adev);
@ -1188,7 +1197,7 @@ static ssize_t amdgpu_set_pp_mclk_od(struct device *dev,
}
if (is_support_sw_smu(adev)) {
value = smu_set_od_percentage(&(adev->smu), OD_MCLK, (uint32_t)value);
value = smu_set_od_percentage(&(adev->smu), SMU_OD_MCLK, (uint32_t)value);
} else {
if (adev->powerplay.pp_funcs->set_mclk_od)
amdgpu_dpm_set_mclk_od(adev, (uint32_t)value);
@ -1453,7 +1462,7 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
struct amdgpu_device *adev = dev_get_drvdata(dev);
struct drm_device *ddev = adev->ddev;
int channel = to_sensor_dev_attr(attr)->index;
int r, temp, size = sizeof(temp);
int r, temp = 0, size = sizeof(temp);
/* Can't get temperature when the card is off */
if ((adev->flags & AMD_IS_PX) &&
@ -2701,6 +2710,44 @@ void amdgpu_pm_print_power_states(struct amdgpu_device *adev)
}
int amdgpu_pm_virt_sysfs_init(struct amdgpu_device *adev)
{
int ret = 0;
if (!(amdgpu_sriov_vf(adev) && amdgim_is_hwperf(adev)))
return ret;
ret = device_create_file(adev->dev, &dev_attr_pp_dpm_sclk);
if (ret) {
DRM_ERROR("failed to create device file pp_dpm_sclk\n");
return ret;
}
ret = device_create_file(adev->dev, &dev_attr_pp_dpm_mclk);
if (ret) {
DRM_ERROR("failed to create device file pp_dpm_mclk\n");
return ret;
}
ret = device_create_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
if (ret) {
DRM_ERROR("failed to create device file for dpm state\n");
return ret;
}
return ret;
}
void amdgpu_pm_virt_sysfs_fini(struct amdgpu_device *adev)
{
if (!(amdgpu_sriov_vf(adev) && amdgim_is_hwperf(adev)))
return;
device_remove_file(adev->dev, &dev_attr_power_dpm_force_performance_level);
device_remove_file(adev->dev, &dev_attr_pp_dpm_sclk);
device_remove_file(adev->dev, &dev_attr_pp_dpm_mclk);
}
int amdgpu_pm_load_smu_firmware(struct amdgpu_device *adev, uint32_t *smu_version)
{
int r;

2
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
Просмотреть файл

@ -32,7 +32,9 @@ struct cg_flag_name
void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
int amdgpu_pm_sysfs_init(struct amdgpu_device *adev);
int amdgpu_pm_virt_sysfs_init(struct amdgpu_device *adev);
void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev);
void amdgpu_pm_virt_sysfs_fini(struct amdgpu_device *adev);
void amdgpu_pm_print_power_states(struct amdgpu_device *adev);
int amdgpu_pm_load_smu_firmware(struct amdgpu_device *adev, uint32_t *smu_version);
void amdgpu_pm_compute_clocks(struct amdgpu_device *adev);

280
drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.c Normal file
Просмотреть файл

@ -0,0 +1,280 @@
/*
* Copyright 2019 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.
*
* Author: Jonathan Kim <jonathan.kim@amd.com>
*
*/
#include <linux/perf_event.h>
#include <linux/init.h>
#include "amdgpu.h"
#include "amdgpu_pmu.h"
#include "df_v3_6.h"
#define PMU_NAME_SIZE 32
/* record to keep track of pmu entry per pmu type per device */
struct amdgpu_pmu_entry {
struct list_head entry;
struct amdgpu_device *adev;
struct pmu pmu;
unsigned int pmu_perf_type;
};
static LIST_HEAD(amdgpu_pmu_list);
/* initialize perf counter */
static int amdgpu_perf_event_init(struct perf_event *event)
{
struct hw_perf_event *hwc = &event->hw;
/* test the event attr type check for PMU enumeration */
if (event->attr.type != event->pmu->type)
return -ENOENT;
/* update the hw_perf_event struct with config data */
hwc->conf = event->attr.config;
return 0;
}
/* start perf counter */
static void amdgpu_perf_start(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
return;
WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
hwc->state = 0;
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
if (!(flags & PERF_EF_RELOAD))
pe->adev->df_funcs->pmc_start(pe->adev, hwc->conf, 1);
pe->adev->df_funcs->pmc_start(pe->adev, hwc->conf, 0);
break;
default:
break;
}
perf_event_update_userpage(event);
}
/* read perf counter */
static void amdgpu_perf_read(struct perf_event *event)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
u64 count, prev;
do {
prev = local64_read(&hwc->prev_count);
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
pe->adev->df_funcs->pmc_get_count(pe->adev, hwc->conf,
&count);
break;
default:
count = 0;
break;
};
} while (local64_cmpxchg(&hwc->prev_count, prev, count) != prev);
local64_add(count - prev, &event->count);
}
/* stop perf counter */
static void amdgpu_perf_stop(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
if (hwc->state & PERF_HES_UPTODATE)
return;
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
pe->adev->df_funcs->pmc_stop(pe->adev, hwc->conf, 0);
break;
default:
break;
};
WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
hwc->state |= PERF_HES_STOPPED;
if (hwc->state & PERF_HES_UPTODATE)
return;
amdgpu_perf_read(event);
hwc->state |= PERF_HES_UPTODATE;
}
/* add perf counter */
static int amdgpu_perf_add(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
int retval;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
retval = pe->adev->df_funcs->pmc_start(pe->adev, hwc->conf, 1);
break;
default:
return 0;
};
if (retval)
return retval;
if (flags & PERF_EF_START)
amdgpu_perf_start(event, PERF_EF_RELOAD);
return retval;
}
/* delete perf counter */
static void amdgpu_perf_del(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
struct amdgpu_pmu_entry *pe = container_of(event->pmu,
struct amdgpu_pmu_entry,
pmu);
amdgpu_perf_stop(event, PERF_EF_UPDATE);
switch (pe->pmu_perf_type) {
case PERF_TYPE_AMDGPU_DF:
pe->adev->df_funcs->pmc_stop(pe->adev, hwc->conf, 1);
break;
default:
break;
};
perf_event_update_userpage(event);
}
/* vega20 pmus */
/* init pmu tracking per pmu type */
static int init_pmu_by_type(struct amdgpu_device *adev,
const struct attribute_group *attr_groups[],
char *pmu_type_name, char *pmu_file_prefix,
unsigned int pmu_perf_type,
unsigned int num_counters)
{
char pmu_name[PMU_NAME_SIZE];
struct amdgpu_pmu_entry *pmu_entry;
int ret = 0;
pmu_entry = kzalloc(sizeof(struct amdgpu_pmu_entry), GFP_KERNEL);
if (!pmu_entry)
return -ENOMEM;
pmu_entry->adev = adev;
pmu_entry->pmu = (struct pmu){
.event_init = amdgpu_perf_event_init,
.add = amdgpu_perf_add,
.del = amdgpu_perf_del,
.start = amdgpu_perf_start,
.stop = amdgpu_perf_stop,
.read = amdgpu_perf_read,
.task_ctx_nr = perf_invalid_context,
};
pmu_entry->pmu.attr_groups = attr_groups;
pmu_entry->pmu_perf_type = pmu_perf_type;
snprintf(pmu_name, PMU_NAME_SIZE, "%s_%d",
pmu_file_prefix, adev->ddev->primary->index);
ret = perf_pmu_register(&pmu_entry->pmu, pmu_name, -1);
if (ret) {
kfree(pmu_entry);
pr_warn("Error initializing AMDGPU %s PMUs.\n", pmu_type_name);
return ret;
}
pr_info("Detected AMDGPU %s Counters. # of Counters = %d.\n",
pmu_type_name, num_counters);
list_add_tail(&pmu_entry->entry, &amdgpu_pmu_list);
return 0;
}
/* init amdgpu_pmu */
int amdgpu_pmu_init(struct amdgpu_device *adev)
{
int ret = 0;
switch (adev->asic_type) {
case CHIP_VEGA20:
/* init df */
ret = init_pmu_by_type(adev, df_v3_6_attr_groups,
"DF", "amdgpu_df", PERF_TYPE_AMDGPU_DF,
DF_V3_6_MAX_COUNTERS);
/* other pmu types go here*/
break;
default:
return 0;
}
return 0;
}
/* destroy all pmu data associated with target device */
void amdgpu_pmu_fini(struct amdgpu_device *adev)
{
struct amdgpu_pmu_entry *pe, *temp;
list_for_each_entry_safe(pe, temp, &amdgpu_pmu_list, entry) {
if (pe->adev == adev) {
list_del(&pe->entry);
perf_pmu_unregister(&pe->pmu);
kfree(pe);
}
}
}

37
drivers/gpu/drm/amd/amdgpu/amdgpu_pmu.h Normal file
Просмотреть файл

@ -0,0 +1,37 @@
/*
* Copyright 2019 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.
*
* Author: Jonathan Kim <jonathan.kim@amd.com>
*
*/
#ifndef _AMDGPU_PMU_H_
#define _AMDGPU_PMU_H_
enum amdgpu_pmu_perf_type {
PERF_TYPE_AMDGPU_DF = 0,
PERF_TYPE_AMDGPU_MAX
};
int amdgpu_pmu_init(struct amdgpu_device *adev);
void amdgpu_pmu_fini(struct amdgpu_device *adev);
#endif /* _AMDGPU_PMU_H_ */

175
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
Просмотреть файл

@ -46,12 +46,19 @@ static int psp_early_init(void *handle)
case CHIP_VEGA10:
case CHIP_VEGA12:
psp_v3_1_set_psp_funcs(psp);
psp->autoload_supported = false;
break;
case CHIP_RAVEN:
psp_v10_0_set_psp_funcs(psp);
psp->autoload_supported = false;
break;
case CHIP_VEGA20:
psp_v11_0_set_psp_funcs(psp);
psp->autoload_supported = false;
break;
case CHIP_NAVI10:
psp_v11_0_set_psp_funcs(psp);
psp->autoload_supported = true;
break;
default:
return -EINVAL;
@ -182,10 +189,44 @@ static void psp_prep_tmr_cmd_buf(struct psp_context *psp,
cmd->cmd.cmd_setup_tmr.buf_size = size;
}
static void psp_prep_load_toc_cmd_buf(struct psp_gfx_cmd_resp *cmd,
uint64_t pri_buf_mc, uint32_t size)
{
cmd->cmd_id = GFX_CMD_ID_LOAD_TOC;
cmd->cmd.cmd_load_toc.toc_phy_addr_lo = lower_32_bits(pri_buf_mc);
cmd->cmd.cmd_load_toc.toc_phy_addr_hi = upper_32_bits(pri_buf_mc);
cmd->cmd.cmd_load_toc.toc_size = size;
}
/* Issue LOAD TOC cmd to PSP to part toc and calculate tmr size needed */
static int psp_load_toc(struct psp_context *psp,
uint32_t *tmr_size)
{
int ret;
struct psp_gfx_cmd_resp *cmd;
cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
if (!cmd)
return -ENOMEM;
/* Copy toc to psp firmware private buffer */
memset(psp->fw_pri_buf, 0, PSP_1_MEG);
memcpy(psp->fw_pri_buf, psp->toc_start_addr, psp->toc_bin_size);
psp_prep_load_toc_cmd_buf(cmd, psp->fw_pri_mc_addr, psp->toc_bin_size);
ret = psp_cmd_submit_buf(psp, NULL, cmd,
psp->fence_buf_mc_addr);
if (!ret)
*tmr_size = psp->cmd_buf_mem->resp.tmr_size;
kfree(cmd);
return ret;
}
/* Set up Trusted Memory Region */
static int psp_tmr_init(struct psp_context *psp)
{
int ret;
int tmr_size;
/*
* According to HW engineer, they prefer the TMR address be "naturally
@ -194,7 +235,21 @@ static int psp_tmr_init(struct psp_context *psp)
* Note: this memory need be reserved till the driver
* uninitializes.
*/
ret = amdgpu_bo_create_kernel(psp->adev, PSP_TMR_SIZE, PSP_TMR_SIZE,
tmr_size = PSP_TMR_SIZE;
/* For ASICs support RLC autoload, psp will parse the toc
* and calculate the total size of TMR needed */
if (psp->toc_start_addr &&
psp->toc_bin_size &&
psp->fw_pri_buf) {
ret = psp_load_toc(psp, &tmr_size);
if (ret) {
DRM_ERROR("Failed to load toc\n");
return ret;
}
}
ret = amdgpu_bo_create_kernel(psp->adev, tmr_size, PSP_TMR_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
@ -210,9 +265,10 @@ static int psp_tmr_load(struct psp_context *psp)
if (!cmd)
return -ENOMEM;
psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr, PSP_TMR_SIZE);
DRM_INFO("reserve 0x%x from 0x%llx for PSP TMR SIZE\n",
PSP_TMR_SIZE, psp->tmr_mc_addr);
psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr,
amdgpu_bo_size(psp->tmr_bo));
DRM_INFO("reserve 0x%lx from 0x%llx for PSP TMR\n",
amdgpu_bo_size(psp->tmr_bo), psp->tmr_mc_addr);
ret = psp_cmd_submit_buf(psp, NULL, cmd,
psp->fence_buf_mc_addr);
@ -726,12 +782,24 @@ static int psp_hw_start(struct psp_context *psp)
return ret;
}
ret = psp_tmr_init(psp);
if (ret) {
DRM_ERROR("PSP tmr init failed!\n");
return ret;
}
ret = psp_tmr_load(psp);
if (ret) {
DRM_ERROR("PSP load tmr failed!\n");
return ret;
}
ret = psp_asd_init(psp);
if (ret) {
DRM_ERROR("PSP asd init failed!\n");
return ret;
}
ret = psp_asd_load(psp);
if (ret) {
DRM_ERROR("PSP load asd failed!\n");
@ -823,6 +891,12 @@ static int psp_get_fw_type(struct amdgpu_firmware_info *ucode,
case AMDGPU_UCODE_ID_DMCU_INTV:
*type = GFX_FW_TYPE_DMCU_ISR;
break;
case AMDGPU_UCODE_ID_VCN0_RAM:
*type = GFX_FW_TYPE_VCN0_RAM;
break;
case AMDGPU_UCODE_ID_VCN1_RAM:
*type = GFX_FW_TYPE_VCN1_RAM;
break;
case AMDGPU_UCODE_ID_MAXIMUM:
default:
return -EINVAL;
@ -851,19 +925,45 @@ static int psp_prep_load_ip_fw_cmd_buf(struct amdgpu_firmware_info *ucode,
return ret;
}
static int psp_execute_np_fw_load(struct psp_context *psp,
struct amdgpu_firmware_info *ucode)
{
int ret = 0;
ret = psp_prep_load_ip_fw_cmd_buf(ucode, psp->cmd);
if (ret)
return ret;
ret = psp_cmd_submit_buf(psp, ucode, psp->cmd,
psp->fence_buf_mc_addr);
return ret;
}
static int psp_np_fw_load(struct psp_context *psp)
{
int i, ret;
struct amdgpu_firmware_info *ucode;
struct amdgpu_device* adev = psp->adev;
if (psp->autoload_supported) {
ucode = &adev->firmware.ucode[AMDGPU_UCODE_ID_SMC];
if (!ucode->fw)
goto out;
ret = psp_execute_np_fw_load(psp, ucode);
if (ret)
return ret;
}
out:
for (i = 0; i < adev->firmware.max_ucodes; i++) {
ucode = &adev->firmware.ucode[i];
if (!ucode->fw)
continue;
if (ucode->ucode_id == AMDGPU_UCODE_ID_SMC &&
psp_smu_reload_quirk(psp))
(psp_smu_reload_quirk(psp) || psp->autoload_supported))
continue;
if (amdgpu_sriov_vf(adev) &&
(ucode->ucode_id == AMDGPU_UCODE_ID_SDMA0
@ -871,16 +971,24 @@ static int psp_np_fw_load(struct psp_context *psp)
|| ucode->ucode_id == AMDGPU_UCODE_ID_RLC_G))
/*skip ucode loading in SRIOV VF */
continue;
if (psp->autoload_supported &&
(ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC1_JT ||
ucode->ucode_id == AMDGPU_UCODE_ID_CP_MEC2_JT))
/* skip mec JT when autoload is enabled */
continue;
ret = psp_prep_load_ip_fw_cmd_buf(ucode, psp->cmd);
if (ret)
return ret;
ret = psp_cmd_submit_buf(psp, ucode, psp->cmd,
psp->fence_buf_mc_addr);
ret = psp_execute_np_fw_load(psp, ucode);
if (ret)
return ret;
/* Start rlc autoload after psp recieved all the gfx firmware */
if (ucode->ucode_id == AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM) {
ret = psp_rlc_autoload(psp);
if (ret) {
DRM_ERROR("Failed to start rlc autoload\n");
return ret;
}
}
#if 0
/* check if firmware loaded sucessfully */
if (!amdgpu_psp_check_fw_loading_status(adev, i))
@ -939,18 +1047,6 @@ static int psp_load_fw(struct amdgpu_device *adev)
goto failed;
}
ret = psp_tmr_init(psp);
if (ret) {
DRM_ERROR("PSP tmr init failed!\n");
goto failed;
}
ret = psp_asd_init(psp);
if (ret) {
DRM_ERROR("PSP asd init failed!\n");
goto failed;
}
skip_memalloc:
ret = psp_hw_start(psp);
if (ret)
@ -1098,6 +1194,39 @@ int psp_gpu_reset(struct amdgpu_device *adev)
return psp_mode1_reset(&adev->psp);
}
int psp_rlc_autoload_start(struct psp_context *psp)
{
int ret;
struct psp_gfx_cmd_resp *cmd;
if (amdgpu_sriov_vf(psp->adev))
return 0;
cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
if (!cmd)
return -ENOMEM;
cmd->cmd_id = GFX_CMD_ID_AUTOLOAD_RLC;
ret = psp_cmd_submit_buf(psp, NULL, cmd,
psp->fence_buf_mc_addr);
kfree(cmd);
return ret;
}
int psp_update_vcn_sram(struct amdgpu_device *adev, int inst_idx,
uint64_t cmd_gpu_addr, int cmd_size)
{
struct amdgpu_firmware_info ucode = {0};
ucode.ucode_id = inst_idx ? AMDGPU_UCODE_ID_VCN1_RAM :
AMDGPU_UCODE_ID_VCN0_RAM;
ucode.mc_addr = cmd_gpu_addr;
ucode.ucode_size = cmd_size;
return psp_execute_np_fw_load(&adev->psp, &ucode);
}
static bool psp_check_fw_loading_status(struct amdgpu_device *adev,
enum AMDGPU_UCODE_ID ucode_type)
{

12
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
Просмотреть файл

@ -101,6 +101,7 @@ struct psp_funcs
int (*ras_trigger_error)(struct psp_context *psp,
struct ta_ras_trigger_error_input *info);
int (*ras_cure_posion)(struct psp_context *psp, uint64_t *mode_ptr);
int (*rlc_autoload_start)(struct psp_context *psp);
};
#define AMDGPU_XGMI_MAX_CONNECTED_NODES 64
@ -154,8 +155,10 @@ struct psp_context
uint32_t sos_feature_version;
uint32_t sys_bin_size;
uint32_t sos_bin_size;
uint32_t toc_bin_size;
uint8_t *sys_start_addr;
uint8_t *sos_start_addr;
uint8_t *toc_start_addr;
/* tmr buffer */
struct amdgpu_bo *tmr_bo;
@ -184,6 +187,8 @@ struct psp_context
/* fence value associated with cmd buffer */
atomic_t fence_value;
/* flag to mark whether gfx fw autoload is supported or not */
bool autoload_supported;
/* xgmi ta firmware and buffer */
const struct firmware *ta_fw;
@ -234,6 +239,8 @@ struct amdgpu_psp_funcs {
#define psp_xgmi_set_topology_info(psp, num_device, topology) \
((psp)->funcs->xgmi_set_topology_info ? \
(psp)->funcs->xgmi_set_topology_info((psp), (num_device), (topology)) : -EINVAL)
#define psp_rlc_autoload(psp) \
((psp)->funcs->rlc_autoload_start ? (psp)->funcs->rlc_autoload_start((psp)) : 0)
#define amdgpu_psp_check_fw_loading_status(adev, i) (adev)->firmware.funcs->check_fw_loading_status((adev), (i))
@ -253,12 +260,17 @@ extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
extern const struct amdgpu_ip_block_version psp_v10_0_ip_block;
int psp_gpu_reset(struct amdgpu_device *adev);
int psp_update_vcn_sram(struct amdgpu_device *adev, int inst_idx,
uint64_t cmd_gpu_addr, int cmd_size);
int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
int psp_ras_enable_features(struct psp_context *psp,
union ta_ras_cmd_input *info, bool enable);
int psp_rlc_autoload_start(struct psp_context *psp);
extern const struct amdgpu_ip_block_version psp_v11_0_ip_block;
int psp_reg_program(struct psp_context *psp, enum psp_reg_prog_id reg,
uint32_t value);

64
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
Просмотреть файл

@ -335,12 +335,13 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, const char __user *
case 2:
ret = amdgpu_ras_reserve_vram(adev,
data.inject.address, PAGE_SIZE, &bo);
/* This address might be used already on failure. In fact we can
* perform an injection in such case.
*/
if (ret)
break;
data.inject.address = amdgpu_bo_gpu_offset(bo);
if (ret) {
/* address was offset, now it is absolute.*/
data.inject.address += adev->gmc.vram_start;
if (data.inject.address > adev->gmc.vram_end)
break;
} else
data.inject.address = amdgpu_bo_gpu_offset(bo);
ret = amdgpu_ras_error_inject(adev, &data.inject);
amdgpu_ras_release_vram(adev, &bo);
break;
@ -969,40 +970,24 @@ static int amdgpu_ras_sysfs_remove_all(struct amdgpu_device *adev)
/* sysfs end */
/* debugfs begin */
static int amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev)
static void amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
struct drm_minor *minor = adev->ddev->primary;
struct dentry *root = minor->debugfs_root, *dir;
struct dentry *ent;
dir = debugfs_create_dir("ras", root);
if (IS_ERR(dir))
return -EINVAL;
con->dir = dir;
ent = debugfs_create_file("ras_ctrl",
S_IWUGO | S_IRUGO, con->dir,
adev, &amdgpu_ras_debugfs_ctrl_ops);
if (IS_ERR(ent)) {
debugfs_remove(con->dir);
return -EINVAL;
}
con->ent = ent;
return 0;
con->dir = debugfs_create_dir("ras", minor->debugfs_root);
con->ent = debugfs_create_file("ras_ctrl", S_IWUGO | S_IRUGO, con->dir,
adev, &amdgpu_ras_debugfs_ctrl_ops);
}
int amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
void amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
struct ras_fs_if *head)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head);
struct dentry *ent;
if (!obj || obj->ent)
return -EINVAL;
return;
get_obj(obj);
@ -1010,34 +995,25 @@ int amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
head->debugfs_name,
sizeof(obj->fs_data.debugfs_name));
ent = debugfs_create_file(obj->fs_data.debugfs_name,
S_IWUGO | S_IRUGO, con->dir,
obj, &amdgpu_ras_debugfs_ops);
if (IS_ERR(ent))
return -EINVAL;
obj->ent = ent;
return 0;
obj->ent = debugfs_create_file(obj->fs_data.debugfs_name,
S_IWUGO | S_IRUGO, con->dir, obj,
&amdgpu_ras_debugfs_ops);
}
int amdgpu_ras_debugfs_remove(struct amdgpu_device *adev,
void amdgpu_ras_debugfs_remove(struct amdgpu_device *adev,
struct ras_common_if *head)
{
struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);
if (!obj || !obj->ent)
return 0;
return;
debugfs_remove(obj->ent);
obj->ent = NULL;
put_obj(obj);
return 0;
}
static int amdgpu_ras_debugfs_remove_all(struct amdgpu_device *adev)
static void amdgpu_ras_debugfs_remove_all(struct amdgpu_device *adev)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
struct ras_manager *obj, *tmp;
@ -1050,8 +1026,6 @@ static int amdgpu_ras_debugfs_remove_all(struct amdgpu_device *adev)
debugfs_remove(con->dir);
con->dir = NULL;
con->ent = NULL;
return 0;
}
/* debugfs end */

4
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h
Просмотреть файл

@ -276,10 +276,10 @@ int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev,
struct ras_common_if *head);
int amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
void amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
struct ras_fs_if *head);
int amdgpu_ras_debugfs_remove(struct amdgpu_device *adev,
void amdgpu_ras_debugfs_remove(struct amdgpu_device *adev,
struct ras_common_if *head);
int amdgpu_ras_error_query(struct amdgpu_device *adev,

12
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
Просмотреть файл

@ -282,6 +282,16 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
return r;
}
r = amdgpu_device_wb_get(adev, &ring->trail_fence_offs);
if (r) {
dev_err(adev->dev,
"(%d) ring trail_fence_offs wb alloc failed\n", r);
return r;
}
ring->trail_fence_gpu_addr =
adev->wb.gpu_addr + (ring->trail_fence_offs * 4);
ring->trail_fence_cpu_addr = &adev->wb.wb[ring->trail_fence_offs];
r = amdgpu_device_wb_get(adev, &ring->cond_exe_offs);
if (r) {
dev_err(adev->dev, "(%d) ring cond_exec_polling wb alloc failed\n", r);
@ -400,7 +410,7 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
{
ktime_t deadline = ktime_add_us(ktime_get(), 10000);
if (!ring->funcs->soft_recovery || !fence)
if (amdgpu_sriov_vf(ring->adev) || !ring->funcs->soft_recovery || !fence)
return false;
atomic_inc(&ring->adev->gpu_reset_counter);

16
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
Просмотреть файл

@ -29,8 +29,8 @@
#include <drm/drm_print.h>
/* max number of rings */
#define AMDGPU_MAX_RINGS 23
#define AMDGPU_MAX_GFX_RINGS 1
#define AMDGPU_MAX_RINGS 24
#define AMDGPU_MAX_GFX_RINGS 2
#define AMDGPU_MAX_COMPUTE_RINGS 8
#define AMDGPU_MAX_VCE_RINGS 3
#define AMDGPU_MAX_UVD_ENC_RINGS 2
@ -172,6 +172,7 @@ struct amdgpu_ring_funcs {
enum drm_sched_priority priority);
/* Try to soft recover the ring to make the fence signal */
void (*soft_recovery)(struct amdgpu_ring *ring, unsigned vmid);
int (*preempt_ib)(struct amdgpu_ring *ring);
};
struct amdgpu_ring {
@ -206,6 +207,10 @@ struct amdgpu_ring {
unsigned fence_offs;
uint64_t current_ctx;
char name[16];
u32 trail_seq;
unsigned trail_fence_offs;
u64 trail_fence_gpu_addr;
volatile u32 *trail_fence_cpu_addr;
unsigned cond_exe_offs;
u64 cond_exe_gpu_addr;
volatile u32 *cond_exe_cpu_addr;
@ -246,6 +251,7 @@ struct amdgpu_ring {
#define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
#define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r))
#define amdgpu_ring_patch_cond_exec(r,o) (r)->funcs->patch_cond_exec((r),(o))
#define amdgpu_ring_preempt_ib(r) (r)->funcs->preempt_ib(r)
int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
@ -266,6 +272,12 @@ void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
struct dma_fence *fence);
static inline void amdgpu_ring_set_preempt_cond_exec(struct amdgpu_ring *ring,
bool cond_exec)
{
*ring->cond_exe_cpu_addr = cond_exec;
}
static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
{
int i = 0;

98
drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
Просмотреть файл

@ -26,6 +26,94 @@
#include "clearstate_defs.h"
/* firmware ID used in rlc toc */
typedef enum _FIRMWARE_ID_ {
FIRMWARE_ID_INVALID = 0,
FIRMWARE_ID_RLC_G_UCODE = 1,
FIRMWARE_ID_RLC_TOC = 2,
FIRMWARE_ID_RLCG_SCRATCH = 3,
FIRMWARE_ID_RLC_SRM_ARAM = 4,
FIRMWARE_ID_RLC_SRM_INDEX_ADDR = 5,
FIRMWARE_ID_RLC_SRM_INDEX_DATA = 6,
FIRMWARE_ID_RLC_P_UCODE = 7,
FIRMWARE_ID_RLC_V_UCODE = 8,
FIRMWARE_ID_RLX6_UCODE = 9,
FIRMWARE_ID_RLX6_DRAM_BOOT = 10,
FIRMWARE_ID_GLOBAL_TAP_DELAYS = 11,
FIRMWARE_ID_SE0_TAP_DELAYS = 12,
FIRMWARE_ID_SE1_TAP_DELAYS = 13,
FIRMWARE_ID_GLOBAL_SE0_SE1_SKEW_DELAYS = 14,
FIRMWARE_ID_SDMA0_UCODE = 15,
FIRMWARE_ID_SDMA0_JT = 16,
FIRMWARE_ID_SDMA1_UCODE = 17,
FIRMWARE_ID_SDMA1_JT = 18,
FIRMWARE_ID_CP_CE = 19,
FIRMWARE_ID_CP_PFP = 20,
FIRMWARE_ID_CP_ME = 21,
FIRMWARE_ID_CP_MEC = 22,
FIRMWARE_ID_CP_MES = 23,
FIRMWARE_ID_MES_STACK = 24,
FIRMWARE_ID_RLC_SRM_DRAM_SR = 25,
FIRMWARE_ID_RLCG_SCRATCH_SR = 26,
FIRMWARE_ID_RLCP_SCRATCH_SR = 27,
FIRMWARE_ID_RLCV_SCRATCH_SR = 28,
FIRMWARE_ID_RLX6_DRAM_SR = 29,
FIRMWARE_ID_SDMA0_PG_CONTEXT = 30,
FIRMWARE_ID_SDMA1_PG_CONTEXT = 31,
FIRMWARE_ID_GLOBAL_MUX_SELECT_RAM = 32,
FIRMWARE_ID_SE0_MUX_SELECT_RAM = 33,
FIRMWARE_ID_SE1_MUX_SELECT_RAM = 34,
FIRMWARE_ID_ACCUM_CTRL_RAM = 35,
FIRMWARE_ID_RLCP_CAM = 36,
FIRMWARE_ID_RLC_SPP_CAM_EXT = 37,
FIRMWARE_ID_MAX = 38,
} FIRMWARE_ID;
typedef struct _RLC_TABLE_OF_CONTENT {
union {
unsigned int DW0;
struct {
unsigned int offset : 25;
unsigned int id : 7;
};
};
union {
unsigned int DW1;
struct {
unsigned int load_at_boot : 1;
unsigned int load_at_vddgfx : 1;
unsigned int load_at_reset : 1;
unsigned int memory_destination : 2;
unsigned int vfflr_image_code : 4;
unsigned int load_mode_direct : 1;
unsigned int save_for_vddgfx : 1;
unsigned int save_for_vfflr : 1;
unsigned int reserved : 1;
unsigned int signed_source : 1;
unsigned int size : 18;
};
};
union {
unsigned int DW2;
struct {
unsigned int indirect_addr_reg : 16;
unsigned int index : 16;
};
};
union {
unsigned int DW3;
struct {
unsigned int indirect_data_reg : 16;
unsigned int indirect_start_offset : 16;
};
};
} RLC_TABLE_OF_CONTENT;
#define RLC_TOC_MAX_SIZE 64
struct amdgpu_rlc_funcs {
bool (*is_rlc_enabled)(struct amdgpu_device *adev);
void (*set_safe_mode)(struct amdgpu_device *adev);
@ -85,6 +173,16 @@ struct amdgpu_rlc {
u8 *save_restore_list_srm;
bool is_rlc_v2_1;
/* for rlc autoload */
struct amdgpu_bo *rlc_autoload_bo;
u64 rlc_autoload_gpu_addr;
void *rlc_autoload_ptr;
/* rlc toc buffer */
struct amdgpu_bo *rlc_toc_bo;
uint64_t rlc_toc_gpu_addr;
void *rlc_toc_buf;
};
void amdgpu_gfx_rlc_enter_safe_mode(struct amdgpu_device *adev);

27
drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.c
Просмотреть файл

@ -24,6 +24,10 @@
#include "amdgpu.h"
#include "amdgpu_sdma.h"
#define AMDGPU_CSA_SDMA_SIZE 64
/* SDMA CSA reside in the 3rd page of CSA */
#define AMDGPU_CSA_SDMA_OFFSET (4096 * 2)
/*
* GPU SDMA IP block helpers function.
*/
@ -56,3 +60,26 @@ int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index)
return -EINVAL;
}
uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring,
unsigned vmid)
{
struct amdgpu_device *adev = ring->adev;
uint64_t csa_mc_addr;
uint32_t index = 0;
int r;
if (vmid == 0 || !amdgpu_mcbp)
return 0;
r = amdgpu_sdma_get_index_from_ring(ring, &index);
if (r || index > 31)
csa_mc_addr = 0;
else
csa_mc_addr = amdgpu_csa_vaddr(adev) +
AMDGPU_CSA_SDMA_OFFSET +
index * AMDGPU_CSA_SDMA_SIZE;
return csa_mc_addr;
}

2
drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h
Просмотреть файл

@ -97,5 +97,5 @@ struct amdgpu_buffer_funcs {
struct amdgpu_sdma_instance *
amdgpu_sdma_get_instance_from_ring(struct amdgpu_ring *ring);
int amdgpu_sdma_get_index_from_ring(struct amdgpu_ring *ring, uint32_t *index);
uint64_t amdgpu_sdma_get_csa_mc_addr(struct amdgpu_ring *ring, unsigned vmid);
#endif

82
drivers/gpu/drm/amd/amdgpu/amdgpu_socbb.h Normal file
Просмотреть файл

@ -0,0 +1,82 @@
/*
* Copyright 2019 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 __AMDGPU_SOCBB_H__
#define __AMDGPU_SOCBB_H__
struct gpu_info_voltage_scaling_v1_0 {
uint32_t state;
uint32_t dscclk_mhz;
uint32_t dcfclk_mhz;
uint32_t socclk_mhz;
uint32_t dram_speed_mts;
uint32_t fabricclk_mhz;
uint32_t dispclk_mhz;
uint32_t phyclk_mhz;
uint32_t dppclk_mhz;
};
struct gpu_info_soc_bounding_box_v1_0 {
uint32_t sr_exit_time_us;
uint32_t sr_enter_plus_exit_time_us;
uint32_t urgent_latency_us;
uint32_t urgent_latency_pixel_data_only_us;
uint32_t urgent_latency_pixel_mixed_with_vm_data_us;
uint32_t urgent_latency_vm_data_only_us;
uint32_t writeback_latency_us;
uint32_t ideal_dram_bw_after_urgent_percent;
uint32_t pct_ideal_dram_sdp_bw_after_urgent_pixel_only; // PercentOfIdealDRAMFabricAndSDPPortBWReceivedAfterUrgLatencyPixelDataOnly
uint32_t pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm;
uint32_t pct_ideal_dram_sdp_bw_after_urgent_vm_only;
uint32_t max_avg_sdp_bw_use_normal_percent;
uint32_t max_avg_dram_bw_use_normal_percent;
uint32_t max_request_size_bytes;
uint32_t downspread_percent;
uint32_t dram_page_open_time_ns;
uint32_t dram_rw_turnaround_time_ns;
uint32_t dram_return_buffer_per_channel_bytes;
uint32_t dram_channel_width_bytes;
uint32_t fabric_datapath_to_dcn_data_return_bytes;
uint32_t dcn_downspread_percent;
uint32_t dispclk_dppclk_vco_speed_mhz;
uint32_t dfs_vco_period_ps;
uint32_t urgent_out_of_order_return_per_channel_pixel_only_bytes;
uint32_t urgent_out_of_order_return_per_channel_pixel_and_vm_bytes;
uint32_t urgent_out_of_order_return_per_channel_vm_only_bytes;
uint32_t round_trip_ping_latency_dcfclk_cycles;
uint32_t urgent_out_of_order_return_per_channel_bytes;
uint32_t channel_interleave_bytes;
uint32_t num_banks;
uint32_t num_chans;
uint32_t vmm_page_size_bytes;
uint32_t dram_clock_change_latency_us;
uint32_t writeback_dram_clock_change_latency_us;
uint32_t return_bus_width_bytes;
uint32_t voltage_override;
uint32_t xfc_bus_transport_time_us;
uint32_t xfc_xbuf_latency_tolerance_us;
uint32_t use_urgent_burst_bw;
uint32_t num_states;
struct gpu_info_voltage_scaling_v1_0 clock_limits[8];
};
#endif

156
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
Просмотреть файл

@ -716,8 +716,7 @@ struct amdgpu_ttm_tt {
struct task_struct *usertask;
uint32_t userflags;
#if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR)
struct hmm_range *ranges;
int nr_ranges;
struct hmm_range *range;
#endif
};
@ -730,57 +729,36 @@ struct amdgpu_ttm_tt {
*/
#if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR)
/* Support Userptr pages cross max 16 vmas */
#define MAX_NR_VMAS (16)
#define MAX_RETRY_HMM_RANGE_FAULT 16
int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
{
struct amdgpu_ttm_tt *gtt = (void *)ttm;
struct mm_struct *mm = gtt->usertask->mm;
unsigned long start = gtt->userptr;
unsigned long end = start + ttm->num_pages * PAGE_SIZE;
struct vm_area_struct *vma = NULL, *vmas[MAX_NR_VMAS];
struct hmm_range *ranges;
unsigned long nr_pages, i;
uint64_t *pfns, f;
struct vm_area_struct *vma;
struct hmm_range *range;
unsigned long i;
uint64_t *pfns;
int retry = 0;
int r = 0;
if (!mm) /* Happens during process shutdown */
return -ESRCH;
down_read(&mm->mmap_sem);
/* user pages may cross multiple VMAs */
gtt->nr_ranges = 0;
do {
unsigned long vm_start;
if (gtt->nr_ranges >= MAX_NR_VMAS) {
DRM_ERROR("Too many VMAs in userptr range\n");
r = -EFAULT;
goto out;
}
vm_start = vma ? vma->vm_end : start;
vma = find_vma(mm, vm_start);
if (unlikely(!vma || vm_start < vma->vm_start)) {
r = -EFAULT;
goto out;
}
vmas[gtt->nr_ranges++] = vma;
} while (end > vma->vm_end);
DRM_DEBUG_DRIVER("0x%lx nr_ranges %d pages 0x%lx\n",
start, gtt->nr_ranges, ttm->num_pages);
vma = find_vma(mm, start);
if (unlikely(!vma || start < vma->vm_start)) {
r = -EFAULT;
goto out;
}
if (unlikely((gtt->userflags & AMDGPU_GEM_USERPTR_ANONONLY) &&
vmas[0]->vm_file)) {
vma->vm_file)) {
r = -EPERM;
goto out;
}
ranges = kvmalloc_array(gtt->nr_ranges, sizeof(*ranges), GFP_KERNEL);
if (unlikely(!ranges)) {
range = kzalloc(sizeof(*range), GFP_KERNEL);
if (unlikely(!range)) {
r = -ENOMEM;
goto out;
}
@ -791,61 +769,67 @@ int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages)
goto out_free_ranges;
}
for (i = 0; i < gtt->nr_ranges; i++)
amdgpu_hmm_init_range(&ranges[i]);
amdgpu_hmm_init_range(range);
range->default_flags = range->flags[HMM_PFN_VALID];
range->default_flags |= amdgpu_ttm_tt_is_readonly(ttm) ?
0 : range->flags[HMM_PFN_WRITE];
range->pfn_flags_mask = 0;
range->pfns = pfns;
hmm_range_register(range, mm, start,
start + ttm->num_pages * PAGE_SIZE, PAGE_SHIFT);
f = ranges[0].flags[HMM_PFN_VALID];
f |= amdgpu_ttm_tt_is_readonly(ttm) ?
0 : ranges[0].flags[HMM_PFN_WRITE];
memset64(pfns, f, ttm->num_pages);
retry:
/*
* Just wait for range to be valid, safe to ignore return value as we
* will use the return value of hmm_range_fault() below under the
* mmap_sem to ascertain the validity of the range.
*/
hmm_range_wait_until_valid(range, HMM_RANGE_DEFAULT_TIMEOUT);
for (nr_pages = 0, i = 0; i < gtt->nr_ranges; i++) {
ranges[i].vma = vmas[i];
ranges[i].start = max(start, vmas[i]->vm_start);
ranges[i].end = min(end, vmas[i]->vm_end);
ranges[i].pfns = pfns + nr_pages;
nr_pages += (ranges[i].end - ranges[i].start) / PAGE_SIZE;
down_read(&mm->mmap_sem);
r = hmm_vma_fault(&ranges[i], true);
if (unlikely(r))
break;
}
if (unlikely(r)) {
while (i--)
hmm_vma_range_done(&ranges[i]);
r = hmm_range_fault(range, true);
if (unlikely(r < 0)) {
if (likely(r == -EAGAIN)) {
/*
* return -EAGAIN, mmap_sem is dropped
*/
if (retry++ < MAX_RETRY_HMM_RANGE_FAULT)
goto retry;
else
pr_err("Retry hmm fault too many times\n");
}
goto out_free_pfns;
goto out_up_read;
}
up_read(&mm->mmap_sem);
for (i = 0; i < ttm->num_pages; i++) {
pages[i] = hmm_pfn_to_page(&ranges[0], pfns[i]);
if (!pages[i]) {
pages[i] = hmm_device_entry_to_page(range, pfns[i]);
if (unlikely(!pages[i])) {
pr_err("Page fault failed for pfn[%lu] = 0x%llx\n",
i, pfns[i]);
goto out_invalid_pfn;
r = -ENOMEM;
goto out_free_pfns;
}
}
gtt->ranges = ranges;
gtt->range = range;
return 0;
out_up_read:
if (likely(r != -EAGAIN))
up_read(&mm->mmap_sem);
out_free_pfns:
hmm_range_unregister(range);
kvfree(pfns);
out_free_ranges:
kvfree(ranges);
kfree(range);
out:
up_read(&mm->mmap_sem);
return r;
out_invalid_pfn:
for (i = 0; i < gtt->nr_ranges; i++)
hmm_vma_range_done(&ranges[i]);
kvfree(pfns);
kvfree(ranges);
return -ENOMEM;
}
/**
@ -858,23 +842,23 @@ bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm)
{
struct amdgpu_ttm_tt *gtt = (void *)ttm;
bool r = false;
int i;
if (!gtt || !gtt->userptr)
return false;
DRM_DEBUG_DRIVER("user_pages_done 0x%llx nr_ranges %d pages 0x%lx\n",
gtt->userptr, gtt->nr_ranges, ttm->num_pages);
DRM_DEBUG_DRIVER("user_pages_done 0x%llx pages 0x%lx\n",
gtt->userptr, ttm->num_pages);
WARN_ONCE(!gtt->ranges || !gtt->ranges[0].pfns,
WARN_ONCE(!gtt->range || !gtt->range->pfns,
"No user pages to check\n");
if (gtt->ranges) {
for (i = 0; i < gtt->nr_ranges; i++)
r |= hmm_vma_range_done(&gtt->ranges[i]);
kvfree(gtt->ranges[0].pfns);
kvfree(gtt->ranges);
gtt->ranges = NULL;
if (gtt->range) {
r = hmm_range_valid(gtt->range);
hmm_range_unregister(gtt->range);
kvfree(gtt->range->pfns);
kfree(gtt->range);
gtt->range = NULL;
}
return r;
@ -958,9 +942,9 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
sg_free_table(ttm->sg);
#if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR)
if (gtt->ranges &&
ttm->pages[0] == hmm_pfn_to_page(&gtt->ranges[0],
gtt->ranges[0].pfns[0]))
if (gtt->range &&
ttm->pages[0] == hmm_device_entry_to_page(gtt->range,
gtt->range->pfns[0]))
WARN_ONCE(1, "Missing get_user_page_done\n");
#endif
}
@ -983,8 +967,8 @@ int amdgpu_ttm_gart_bind(struct amdgpu_device *adev,
goto gart_bind_fail;
/* Patch mtype of the second part BO */
flags &= ~AMDGPU_PTE_MTYPE_MASK;
flags |= AMDGPU_PTE_MTYPE(AMDGPU_MTYPE_NC);
flags &= ~AMDGPU_PTE_MTYPE_VG10_MASK;
flags |= AMDGPU_PTE_MTYPE_VG10(AMDGPU_MTYPE_NC);
r = amdgpu_gart_bind(adev,
gtt->offset + (page_idx << PAGE_SHIFT),

43
drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
Просмотреть файл

@ -77,6 +77,14 @@ void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr)
container_of(hdr, struct smc_firmware_header_v1_0, header);
DRM_DEBUG("ucode_start_addr: %u\n", le32_to_cpu(smc_hdr->ucode_start_addr));
} else if (version_major == 2) {
const struct smc_firmware_header_v1_0 *v1_hdr =
container_of(hdr, struct smc_firmware_header_v1_0, header);
const struct smc_firmware_header_v2_0 *v2_hdr =
container_of(v1_hdr, struct smc_firmware_header_v2_0, v1_0);
DRM_INFO("ppt_offset_bytes: %u\n", le32_to_cpu(v2_hdr->ppt_offset_bytes));
DRM_INFO("ppt_size_bytes: %u\n", le32_to_cpu(v2_hdr->ppt_size_bytes));
} else {
DRM_ERROR("Unknown SMC ucode version: %u.%u\n", version_major, version_minor);
}
@ -227,6 +235,40 @@ void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr)
}
}
void amdgpu_ucode_print_psp_hdr(const struct common_firmware_header *hdr)
{
uint16_t version_major = le16_to_cpu(hdr->header_version_major);
uint16_t version_minor = le16_to_cpu(hdr->header_version_minor);
DRM_DEBUG("PSP\n");
amdgpu_ucode_print_common_hdr(hdr);
if (version_major == 1) {
const struct psp_firmware_header_v1_0 *psp_hdr =
container_of(hdr, struct psp_firmware_header_v1_0, header);
DRM_DEBUG("ucode_feature_version: %u\n",
le32_to_cpu(psp_hdr->ucode_feature_version));
DRM_DEBUG("sos_offset_bytes: %u\n",
le32_to_cpu(psp_hdr->sos_offset_bytes));
DRM_DEBUG("sos_size_bytes: %u\n",
le32_to_cpu(psp_hdr->sos_size_bytes));
if (version_minor == 1) {
const struct psp_firmware_header_v1_1 *psp_hdr_v1_1 =
container_of(psp_hdr, struct psp_firmware_header_v1_1, v1_0);
DRM_DEBUG("toc_header_version: %u\n",
le32_to_cpu(psp_hdr_v1_1->toc_header_version));
DRM_DEBUG("toc_offset_bytes: %u\n",
le32_to_cpu(psp_hdr_v1_1->toc_offset_bytes));
DRM_DEBUG("toc_size_bytes: %u\n",
le32_to_cpu(psp_hdr_v1_1->toc_size_bytes));
}
} else {
DRM_ERROR("Unknown PSP ucode version: %u.%u\n",
version_major, version_minor);
}
}
void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr)
{
uint16_t version_major = le16_to_cpu(hdr->header_version_major);
@ -302,6 +344,7 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type)
case CHIP_RAVEN:
case CHIP_VEGA12:
case CHIP_VEGA20:
case CHIP_NAVI10:
if (!load_type)
return AMDGPU_FW_LOAD_DIRECT;
else

66
drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
Просмотреть файл

@ -23,6 +23,8 @@
#ifndef __AMDGPU_UCODE_H__
#define __AMDGPU_UCODE_H__
#include "amdgpu_socbb.h"
struct common_firmware_header {
uint32_t size_bytes; /* size of the entire header+image(s) in bytes */
uint32_t header_size_bytes; /* size of just the header in bytes */
@ -49,6 +51,26 @@ struct smc_firmware_header_v1_0 {
uint32_t ucode_start_addr;
};
/* version_major=2, version_minor=0 */
struct smc_firmware_header_v2_0 {
struct smc_firmware_header_v1_0 v1_0;
uint32_t ppt_offset_bytes; /* soft pptable offset */
uint32_t ppt_size_bytes; /* soft pptable size */
};
struct smc_soft_pptable_entry {
uint32_t id;
uint32_t ppt_offset_bytes;
uint32_t ppt_size_bytes;
};
/* version_major=2, version_minor=1 */
struct smc_firmware_header_v2_1 {
struct smc_firmware_header_v1_0 v1_0;
uint32_t pptable_count;
uint32_t pptable_entry_offset;
};
/* version_major=1, version_minor=0 */
struct psp_firmware_header_v1_0 {
struct common_firmware_header header;
@ -57,6 +79,14 @@ struct psp_firmware_header_v1_0 {
uint32_t sos_size_bytes;
};
/* version_major=1, version_minor=1 */
struct psp_firmware_header_v1_1 {
struct psp_firmware_header_v1_0 v1_0;
uint32_t toc_header_version;
uint32_t toc_offset_bytes;
uint32_t toc_size_bytes;
};
/* version_major=1, version_minor=0 */
struct ta_firmware_header_v1_0 {
struct common_firmware_header header;
@ -76,6 +106,21 @@ struct gfx_firmware_header_v1_0 {
uint32_t jt_size; /* size of jt */
};
/* version_major=1, version_minor=0 */
struct mes_firmware_header_v1_0 {
struct common_firmware_header header;
uint32_t mes_ucode_version;
uint32_t mes_ucode_size_bytes;
uint32_t mes_ucode_offset_bytes;
uint32_t mes_ucode_data_version;
uint32_t mes_ucode_data_size_bytes;
uint32_t mes_ucode_data_offset_bytes;
uint32_t mes_uc_start_addr_lo;
uint32_t mes_uc_start_addr_hi;
uint32_t mes_data_start_addr_lo;
uint32_t mes_data_start_addr_hi;
};
/* version_major=1, version_minor=0 */
struct rlc_firmware_header_v1_0 {
struct common_firmware_header header;
@ -161,6 +206,19 @@ struct gpu_info_firmware_v1_0 {
uint32_t gc_lds_size;
};
struct gpu_info_firmware_v1_1 {
struct gpu_info_firmware_v1_0 v1_0;
uint32_t num_sc_per_sh;
uint32_t num_packer_per_sc;
};
/* gpu info payload
* version_major=1, version_minor=1 */
struct gpu_info_firmware_v1_2 {
struct gpu_info_firmware_v1_1 v1_1;
struct gpu_info_soc_bounding_box_v1_0 soc_bounding_box;
};
/* version_major=1, version_minor=0 */
struct gpu_info_firmware_header_v1_0 {
struct common_firmware_header header;
@ -180,7 +238,9 @@ union amdgpu_firmware_header {
struct common_firmware_header common;
struct mc_firmware_header_v1_0 mc;
struct smc_firmware_header_v1_0 smc;
struct smc_firmware_header_v2_0 smc_v2_0;
struct psp_firmware_header_v1_0 psp;
struct psp_firmware_header_v1_1 psp_v1_1;
struct ta_firmware_header_v1_0 ta;
struct gfx_firmware_header_v1_0 gfx;
struct rlc_firmware_header_v1_0 rlc;
@ -206,6 +266,8 @@ enum AMDGPU_UCODE_ID {
AMDGPU_UCODE_ID_CP_MEC1_JT,
AMDGPU_UCODE_ID_CP_MEC2,
AMDGPU_UCODE_ID_CP_MEC2_JT,
AMDGPU_UCODE_ID_CP_MES,
AMDGPU_UCODE_ID_CP_MES_DATA,
AMDGPU_UCODE_ID_RLC_G,
AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL,
AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM,
@ -218,6 +280,8 @@ enum AMDGPU_UCODE_ID {
AMDGPU_UCODE_ID_VCN,
AMDGPU_UCODE_ID_DMCU_ERAM,
AMDGPU_UCODE_ID_DMCU_INTV,
AMDGPU_UCODE_ID_VCN0_RAM,
AMDGPU_UCODE_ID_VCN1_RAM,
AMDGPU_UCODE_ID_MAXIMUM,
};
@ -232,6 +296,7 @@ enum amdgpu_firmware_load_type {
AMDGPU_FW_LOAD_DIRECT = 0,
AMDGPU_FW_LOAD_SMU,
AMDGPU_FW_LOAD_PSP,
AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO,
};
/* conform to smu_ucode_xfer_cz.h */
@ -284,6 +349,7 @@ void amdgpu_ucode_print_smc_hdr(const struct common_firmware_header *hdr);
void amdgpu_ucode_print_gfx_hdr(const struct common_firmware_header *hdr);
void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr);
void amdgpu_ucode_print_sdma_hdr(const struct common_firmware_header *hdr);
void amdgpu_ucode_print_psp_hdr(const struct common_firmware_header *hdr);
void amdgpu_ucode_print_gpu_info_hdr(const struct common_firmware_header *hdr);
int amdgpu_ucode_validate(const struct firmware *fw);
bool amdgpu_ucode_hdr_version(union amdgpu_firmware_header *hdr,

56
drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
Просмотреть файл

@ -46,10 +46,12 @@
#define FIRMWARE_RAVEN "amdgpu/raven_vcn.bin"
#define FIRMWARE_PICASSO "amdgpu/picasso_vcn.bin"
#define FIRMWARE_RAVEN2 "amdgpu/raven2_vcn.bin"
#define FIRMWARE_NAVI10 "amdgpu/navi10_vcn.bin"
MODULE_FIRMWARE(FIRMWARE_RAVEN);
MODULE_FIRMWARE(FIRMWARE_PICASSO);
MODULE_FIRMWARE(FIRMWARE_RAVEN2);
MODULE_FIRMWARE(FIRMWARE_NAVI10);
static void amdgpu_vcn_idle_work_handler(struct work_struct *work);
@ -72,6 +74,12 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
else
fw_name = FIRMWARE_RAVEN;
break;
case CHIP_NAVI10:
fw_name = FIRMWARE_NAVI10;
if ((adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) &&
(adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG))
adev->vcn.indirect_sram = true;
break;
default:
return -EINVAL;
}
@ -133,6 +141,16 @@ int amdgpu_vcn_sw_init(struct amdgpu_device *adev)
return r;
}
if (adev->vcn.indirect_sram) {
r = amdgpu_bo_create_kernel(adev, 64 * 2 * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM, &adev->vcn.dpg_sram_bo,
&adev->vcn.dpg_sram_gpu_addr, &adev->vcn.dpg_sram_cpu_addr);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate DPG bo\n", r);
return r;
}
}
return 0;
}
@ -142,6 +160,12 @@ int amdgpu_vcn_sw_fini(struct amdgpu_device *adev)
kvfree(adev->vcn.saved_bo);
if (adev->vcn.indirect_sram) {
amdgpu_bo_free_kernel(&adev->vcn.dpg_sram_bo,
&adev->vcn.dpg_sram_gpu_addr,
(void **)&adev->vcn.dpg_sram_cpu_addr);
}
amdgpu_bo_free_kernel(&adev->vcn.vcpu_bo,
&adev->vcn.gpu_addr,
(void **)&adev->vcn.cpu_addr);
@ -245,7 +269,7 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
if (fences == 0) {
amdgpu_gfx_off_ctrl(adev, true);
if (adev->pm.dpm_enabled)
if (adev->asic_type < CHIP_NAVI10 && adev->pm.dpm_enabled)
amdgpu_dpm_enable_uvd(adev, false);
else
amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
@ -262,7 +286,7 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
if (set_clocks) {
amdgpu_gfx_off_ctrl(adev, false);
if (adev->pm.dpm_enabled)
if (adev->asic_type < CHIP_NAVI10 && adev->pm.dpm_enabled)
amdgpu_dpm_enable_uvd(adev, true);
else
amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCN,
@ -308,17 +332,15 @@ int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring)
unsigned i;
int r;
WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0xCAFEDEAD);
WREG32(adev->vcn.external.scratch9, 0xCAFEDEAD);
r = amdgpu_ring_alloc(ring, 3);
if (r)
return r;
amdgpu_ring_write(ring,
PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0));
amdgpu_ring_write(ring, PACKET0(adev->vcn.internal.scratch9, 0));
amdgpu_ring_write(ring, 0xDEADBEEF);
amdgpu_ring_commit(ring);
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9));
tmp = RREG32(adev->vcn.external.scratch9);
if (tmp == 0xDEADBEEF)
break;
udelay(1);
@ -347,14 +369,14 @@ static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
ib = &job->ibs[0];
addr = amdgpu_bo_gpu_offset(bo);
ib->ptr[0] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0), 0);
ib->ptr[0] = PACKET0(adev->vcn.internal.data0, 0);
ib->ptr[1] = addr;
ib->ptr[2] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1), 0);
ib->ptr[2] = PACKET0(adev->vcn.internal.data1, 0);
ib->ptr[3] = addr >> 32;
ib->ptr[4] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD), 0);
ib->ptr[4] = PACKET0(adev->vcn.internal.cmd, 0);
ib->ptr[5] = 0;
for (i = 6; i < 16; i += 2) {
ib->ptr[i] = PACKET0(SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP), 0);
ib->ptr[i] = PACKET0(adev->vcn.internal.nop, 0);
ib->ptr[i+1] = 0;
}
ib->length_dw = 16;
@ -629,19 +651,17 @@ int amdgpu_vcn_jpeg_ring_test_ring(struct amdgpu_ring *ring)
unsigned i;
int r;
WREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0xCAFEDEAD);
WREG32(adev->vcn.external.jpeg_pitch, 0xCAFEDEAD);
r = amdgpu_ring_alloc(ring, 3);
if (r)
return r;
amdgpu_ring_write(ring,
PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0, 0, 0));
amdgpu_ring_write(ring, PACKET0(adev->vcn.internal.jpeg_pitch, 0));
amdgpu_ring_write(ring, 0xDEADBEEF);
amdgpu_ring_commit(ring);
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9));
tmp = RREG32(adev->vcn.external.jpeg_pitch);
if (tmp == 0xDEADBEEF)
break;
udelay(1);
@ -669,7 +689,7 @@ static int amdgpu_vcn_jpeg_set_reg(struct amdgpu_ring *ring, uint32_t handle,
ib = &job->ibs[0];
ib->ptr[0] = PACKETJ(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9), 0, 0, PACKETJ_TYPE0);
ib->ptr[0] = PACKETJ(adev->vcn.internal.jpeg_pitch, 0, 0, PACKETJ_TYPE0);
ib->ptr[1] = 0xDEADBEEF;
for (i = 2; i < 16; i += 2) {
ib->ptr[i] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
@ -715,7 +735,7 @@ int amdgpu_vcn_jpeg_ring_test_ib(struct amdgpu_ring *ring, long timeout)
}
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32(SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9));
tmp = RREG32(adev->vcn.external.jpeg_pitch);
if (tmp == 0xDEADBEEF)
break;
udelay(1);

71
drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h
Просмотреть файл

@ -25,7 +25,7 @@
#define __AMDGPU_VCN_H__
#define AMDGPU_VCN_STACK_SIZE (128*1024)
#define AMDGPU_VCN_CONTEXT_SIZE (512*1024)
#define AMDGPU_VCN_CONTEXT_SIZE (512*1024)
#define AMDGPU_VCN_FIRMWARE_OFFSET 256
#define AMDGPU_VCN_MAX_ENC_RINGS 3
@ -45,6 +45,11 @@
#define VCN_ENC_CMD_REG_WRITE 0x0000000b
#define VCN_ENC_CMD_REG_WAIT 0x0000000c
#define VCN_VID_SOC_ADDRESS_2_0 0x1fa00
#define VCN_AON_SOC_ADDRESS_2_0 0x1f800
#define VCN_VID_IP_ADDRESS_2_0 0x0
#define VCN_AON_IP_ADDRESS_2_0 0x30000
#define RREG32_SOC15_DPG_MODE(ip, inst, reg, mask, sram_sel) \
({ WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_MASK, mask); \
WREG32_SOC15(ip, inst, mmUVD_DPG_LMA_CTL, \
@ -66,8 +71,55 @@
(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); \
} while (0)
#define SOC15_DPG_MODE_OFFSET_2_0(ip, inst, reg) \
({ \
uint32_t internal_reg_offset, addr; \
bool video_range, aon_range; \
\
addr = (adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg); \
addr <<= 2; \
video_range = ((((0xFFFFF & addr) >= (VCN_VID_SOC_ADDRESS_2_0)) && \
((0xFFFFF & addr) < ((VCN_VID_SOC_ADDRESS_2_0 + 0x2600))))); \
aon_range = ((((0xFFFFF & addr) >= (VCN_AON_SOC_ADDRESS_2_0)) && \
((0xFFFFF & addr) < ((VCN_AON_SOC_ADDRESS_2_0 + 0x600))))); \
if (video_range) \
internal_reg_offset = ((0xFFFFF & addr) - (VCN_VID_SOC_ADDRESS_2_0) + \
(VCN_VID_IP_ADDRESS_2_0)); \
else if (aon_range) \
internal_reg_offset = ((0xFFFFF & addr) - (VCN_AON_SOC_ADDRESS_2_0) + \
(VCN_AON_IP_ADDRESS_2_0)); \
else \
internal_reg_offset = (0xFFFFF & addr); \
\
internal_reg_offset >>= 2; \
})
#define RREG32_SOC15_DPG_MODE_2_0(offset, mask_en) \
({ \
WREG32_SOC15(VCN, 0, mmUVD_DPG_LMA_CTL, \
(0x0 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT | \
mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT | \
offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT)); \
RREG32_SOC15(VCN, 0, mmUVD_DPG_LMA_DATA); \
})
#define WREG32_SOC15_DPG_MODE_2_0(offset, value, mask_en, indirect) \
do { \
if (!indirect) { \
WREG32_SOC15(VCN, 0, mmUVD_DPG_LMA_DATA, value); \
WREG32_SOC15(VCN, 0, mmUVD_DPG_LMA_CTL, \
(0x1 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT | \
mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT | \
offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT)); \
} else { \
*adev->vcn.dpg_sram_curr_addr++ = offset; \
*adev->vcn.dpg_sram_curr_addr++ = value; \
} \
} while (0)
enum engine_status_constants {
UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON = 0x2AAAA0,
UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON_2_0 = 0xAAAA0,
UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON = 0x00000002,
UVD_STATUS__UVD_BUSY = 0x00000004,
GB_ADDR_CONFIG_DEFAULT = 0x26010011,
@ -75,6 +127,7 @@ enum engine_status_constants {
UVD_STATUS__BUSY = 0x5,
UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF = 0x1,
UVD_STATUS__RBC_BUSY = 0x1,
UVD_PGFSM_STATUS_UVDJ_PWR_ON = 0,
};
enum internal_dpg_state {
@ -87,6 +140,15 @@ struct dpg_pause_state {
enum internal_dpg_state jpeg;
};
struct amdgpu_vcn_reg{
unsigned data0;
unsigned data1;
unsigned cmd;
unsigned nop;
unsigned scratch9;
unsigned jpeg_pitch;
};
struct amdgpu_vcn {
struct amdgpu_bo *vcpu_bo;
void *cpu_addr;
@ -102,8 +164,15 @@ struct amdgpu_vcn {
unsigned num_enc_rings;
enum amd_powergating_state cur_state;
struct dpg_pause_state pause_state;
struct amdgpu_vcn_reg internal, external;
int (*pause_dpg_mode)(struct amdgpu_device *adev,
struct dpg_pause_state *new_state);
bool indirect_sram;
struct amdgpu_bo *dpg_sram_bo;
void *dpg_sram_cpu_addr;
uint64_t dpg_sram_gpu_addr;
uint32_t *dpg_sram_curr_addr;
};
int amdgpu_vcn_sw_init(struct amdgpu_device *adev);

1
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
Просмотреть файл

@ -253,6 +253,7 @@ typedef struct amdgim_vf2pf_info_v2 amdgim_vf2pf_info ;
struct amdgpu_virt {
uint32_t caps;
struct amdgpu_bo *csa_obj;
void *csa_cpu_addr;
bool chained_ib_support;
uint32_t reg_val_offs;
struct amdgpu_irq_src ack_irq;

14
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
Просмотреть файл

@ -1574,12 +1574,22 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
flags &= ~AMDGPU_PTE_EXECUTABLE;
flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE;
flags &= ~AMDGPU_PTE_MTYPE_MASK;
flags |= (mapping->flags & AMDGPU_PTE_MTYPE_MASK);
if (adev->asic_type == CHIP_NAVI10) {
flags &= ~AMDGPU_PTE_MTYPE_NV10_MASK;
flags |= (mapping->flags & AMDGPU_PTE_MTYPE_NV10_MASK);
} else {
flags &= ~AMDGPU_PTE_MTYPE_VG10_MASK;
flags |= (mapping->flags & AMDGPU_PTE_MTYPE_VG10_MASK);
}
if ((mapping->flags & AMDGPU_PTE_PRT) &&
(adev->asic_type >= CHIP_VEGA10)) {
flags |= AMDGPU_PTE_PRT;
if (adev->asic_type >= CHIP_NAVI10) {
flags |= AMDGPU_PTE_SNOOPED;
flags |= AMDGPU_PTE_LOG;
flags |= AMDGPU_PTE_SYSTEM;
}
flags &= ~AMDGPU_PTE_VALID;
}

12
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
Просмотреть файл

@ -67,6 +67,8 @@ struct amdgpu_bo_list_entry;
/* PDE is handled as PTE for VEGA10 */
#define AMDGPU_PDE_PTE (1ULL << 54)
#define AMDGPU_PTE_LOG (1ULL << 55)
/* PTE is handled as PDE for VEGA10 (Translate Further) */
#define AMDGPU_PTE_TF (1ULL << 56)
@ -75,8 +77,8 @@ struct amdgpu_bo_list_entry;
/* For GFX9 */
#define AMDGPU_PTE_MTYPE(a) ((uint64_t)a << 57)
#define AMDGPU_PTE_MTYPE_MASK AMDGPU_PTE_MTYPE(3ULL)
#define AMDGPU_PTE_MTYPE_VG10(a) ((uint64_t)(a) << 57)
#define AMDGPU_PTE_MTYPE_VG10_MASK AMDGPU_PTE_MTYPE_VG10(3ULL)
#define AMDGPU_MTYPE_NC 0
#define AMDGPU_MTYPE_CC 2
@ -86,7 +88,11 @@ struct amdgpu_bo_list_entry;
| AMDGPU_PTE_EXECUTABLE \
| AMDGPU_PTE_READABLE \
| AMDGPU_PTE_WRITEABLE \
| AMDGPU_PTE_MTYPE(AMDGPU_MTYPE_CC))
| AMDGPU_PTE_MTYPE_VG10(AMDGPU_MTYPE_CC))
/* NAVI10 only */
#define AMDGPU_PTE_MTYPE_NV10(a) ((uint64_t)(a) << 48)
#define AMDGPU_PTE_MTYPE_NV10_MASK AMDGPU_PTE_MTYPE_NV10(7ULL)
/* How to programm VM fault handling */
#define AMDGPU_VM_FAULT_STOP_NEVER 0

32
drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
Просмотреть файл

@ -275,7 +275,7 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
struct drm_mm_node *nodes;
enum drm_mm_insert_mode mode;
unsigned long lpfn, num_nodes, pages_per_node, pages_left;
uint64_t usage = 0, vis_usage = 0;
uint64_t vis_usage = 0, mem_bytes;
unsigned i;
int r;
@ -283,20 +283,34 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
if (!lpfn)
lpfn = man->size;
if (place->flags & TTM_PL_FLAG_CONTIGUOUS ||
amdgpu_vram_page_split == -1) {
/* bail out quickly if there's likely not enough VRAM for this BO */
mem_bytes = (u64)mem->num_pages << PAGE_SHIFT;
if (atomic64_add_return(mem_bytes, &mgr->usage) > adev->gmc.mc_vram_size) {
atomic64_sub(mem_bytes, &mgr->usage);
mem->mm_node = NULL;
return 0;
}
if (place->flags & TTM_PL_FLAG_CONTIGUOUS) {
pages_per_node = ~0ul;
num_nodes = 1;
} else {
pages_per_node = max((uint32_t)amdgpu_vram_page_split,
mem->page_alignment);
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
pages_per_node = HPAGE_PMD_NR;
#else
/* default to 2MB */
pages_per_node = (2UL << (20UL - PAGE_SHIFT));
#endif
pages_per_node = max((uint32_t)pages_per_node, mem->page_alignment);
num_nodes = DIV_ROUND_UP(mem->num_pages, pages_per_node);
}
nodes = kvmalloc_array(num_nodes, sizeof(*nodes),
nodes = kvmalloc_array((uint32_t)num_nodes, sizeof(*nodes),
GFP_KERNEL | __GFP_ZERO);
if (!nodes)
if (!nodes) {
atomic64_sub(mem_bytes, &mgr->usage);
return -ENOMEM;
}
mode = DRM_MM_INSERT_BEST;
if (place->flags & TTM_PL_FLAG_TOPDOWN)
@ -316,7 +330,6 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
if (unlikely(r))
break;
usage += nodes[i].size << PAGE_SHIFT;
vis_usage += amdgpu_vram_mgr_vis_size(adev, &nodes[i]);
amdgpu_vram_mgr_virt_start(mem, &nodes[i]);
pages_left -= pages;
@ -336,14 +349,12 @@ static int amdgpu_vram_mgr_new(struct ttm_mem_type_manager *man,
if (unlikely(r))
goto error;
usage += nodes[i].size << PAGE_SHIFT;
vis_usage += amdgpu_vram_mgr_vis_size(adev, &nodes[i]);
amdgpu_vram_mgr_virt_start(mem, &nodes[i]);
pages_left -= pages;
}
spin_unlock(&mgr->lock);
atomic64_add(usage, &mgr->usage);
atomic64_add(vis_usage, &mgr->vis_usage);
mem->mm_node = nodes;
@ -354,6 +365,7 @@ error:
while (i--)
drm_mm_remove_node(&nodes[i]);
spin_unlock(&mgr->lock);
atomic64_sub(mem->num_pages << PAGE_SHIFT, &mgr->usage);
kvfree(nodes);
return r == -ENOSPC ? 0 : r;

101
drivers/gpu/drm/amd/amdgpu/athub_v2_0.c Normal file
Просмотреть файл

@ -0,0 +1,101 @@
/*
* Copyright 2019 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.
*
*/
#include "amdgpu.h"
#include "athub_v2_0.h"
#include "athub/athub_2_0_0_offset.h"
#include "athub/athub_2_0_0_sh_mask.h"
#include "athub/athub_2_0_0_default.h"
#include "navi10_enum.h"
#include "soc15_common.h"
static void
athub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
bool enable)
{
uint32_t def, data;
def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG))
data |= ATHUB_MISC_CNTL__CG_ENABLE_MASK;
else
data &= ~ATHUB_MISC_CNTL__CG_ENABLE_MASK;
if (def != data)
WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
}
static void
athub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
bool enable)
{
uint32_t def, data;
def = data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS) &&
(adev->cg_flags & AMD_CG_SUPPORT_HDP_LS))
data |= ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
else
data &= ~ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK;
if (def != data)
WREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL, data);
}
int athub_v2_0_set_clockgating(struct amdgpu_device *adev,
enum amd_clockgating_state state)
{
if (amdgpu_sriov_vf(adev))
return 0;
switch (adev->asic_type) {
case CHIP_NAVI10:
athub_v2_0_update_medium_grain_clock_gating(adev,
state == AMD_CG_STATE_GATE ? true : false);
athub_v2_0_update_medium_grain_light_sleep(adev,
state == AMD_CG_STATE_GATE ? true : false);
break;
default:
break;
}
return 0;
}
void athub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags)
{
int data;
/* AMD_CG_SUPPORT_ATHUB_MGCG */
data = RREG32_SOC15(ATHUB, 0, mmATHUB_MISC_CNTL);
if (data & ATHUB_MISC_CNTL__CG_ENABLE_MASK)
*flags |= AMD_CG_SUPPORT_ATHUB_MGCG;
/* AMD_CG_SUPPORT_ATHUB_LS */
if (data & ATHUB_MISC_CNTL__CG_MEM_LS_ENABLE_MASK)
*flags |= AMD_CG_SUPPORT_ATHUB_LS;
}

30
drivers/gpu/drm/amd/amdgpu/athub_v2_0.h Normal file
Просмотреть файл

@ -0,0 +1,30 @@
/*
* Copyright 2019 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 __ATHUB_V2_0_H__
#define __ATHUB_V2_0_H__
int athub_v2_0_set_clockgating(struct amdgpu_device *adev,
enum amd_clockgating_state state);
void athub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags);
#endif

975
drivers/gpu/drm/amd/amdgpu/clearstate_gfx10.h Normal file
Просмотреть файл

@ -0,0 +1,975 @@
/*
* Copyright 2019 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.
*
*/
static const unsigned int gfx10_SECT_CONTEXT_def_1[] = {
0x00000000, // DB_RENDER_CONTROL
0x00000000, // DB_COUNT_CONTROL
0x00000000, // DB_DEPTH_VIEW
0x00000000, // DB_RENDER_OVERRIDE
0x00000000, // DB_RENDER_OVERRIDE2
0x00000000, // DB_HTILE_DATA_BASE
0x00000000, // HOLE
0x00000000, // DB_DEPTH_SIZE_XY
0x00000000, // DB_DEPTH_BOUNDS_MIN
0x00000000, // DB_DEPTH_BOUNDS_MAX
0x00000000, // DB_STENCIL_CLEAR
0x00000000, // DB_DEPTH_CLEAR
0x00000000, // PA_SC_SCREEN_SCISSOR_TL
0x40004000, // PA_SC_SCREEN_SCISSOR_BR
0x00000000, // DB_DFSM_CONTROL
0x00000000, // DB_DEPTH_INFO
0x00000000, // DB_Z_INFO
0x00000000, // DB_STENCIL_INFO
0x00000000, // DB_Z_READ_BASE
0x00000000, // DB_STENCIL_READ_BASE
0x00000000, // DB_Z_WRITE_BASE
0x00000000, // DB_STENCIL_WRITE_BASE
0x00000000, // DB_DEPTH_SIZE
0x00000000, // DB_DEPTH_SLICE
0x00000000, // DB_Z_INFO2
0x00000000, // DB_STENCIL_INFO2
0x00000000, // DB_Z_READ_BASE_HI
0x00000000, // DB_STENCIL_READ_BASE_HI
0x00000000, // DB_Z_WRITE_BASE_HI
0x00000000, // DB_STENCIL_WRITE_BASE_HI
0x00000000, // DB_HTILE_DATA_BASE_HI
0x00150055, // DB_RMI_L2_CACHE_CONTROL
0x00000000, // TA_BC_BASE_ADDR
0x00000000, // TA_BC_BASE_ADDR_HI
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // COHER_DEST_BASE_HI_0
0x00000000, // COHER_DEST_BASE_HI_1
0x00000000, // COHER_DEST_BASE_HI_2
0x00000000, // COHER_DEST_BASE_HI_3
0x00000000, // COHER_DEST_BASE_2
0x00000000, // COHER_DEST_BASE_3
0x00000000, // PA_SC_WINDOW_OFFSET
0x80000000, // PA_SC_WINDOW_SCISSOR_TL
0x40004000, // PA_SC_WINDOW_SCISSOR_BR
0x0000ffff, // PA_SC_CLIPRECT_RULE
0x00000000, // PA_SC_CLIPRECT_0_TL
0x40004000, // PA_SC_CLIPRECT_0_BR
0x00000000, // PA_SC_CLIPRECT_1_TL
0x40004000, // PA_SC_CLIPRECT_1_BR
0x00000000, // PA_SC_CLIPRECT_2_TL
0x40004000, // PA_SC_CLIPRECT_2_BR
0x00000000, // PA_SC_CLIPRECT_3_TL
0x40004000, // PA_SC_CLIPRECT_3_BR
0xaa99aaaa, // PA_SC_EDGERULE
0x00000000, // PA_SU_HARDWARE_SCREEN_OFFSET
0xffffffff, // CB_TARGET_MASK
0xffffffff, // CB_SHADER_MASK
0x80000000, // PA_SC_GENERIC_SCISSOR_TL
0x40004000, // PA_SC_GENERIC_SCISSOR_BR
0x00000000, // COHER_DEST_BASE_0
0x00000000, // COHER_DEST_BASE_1
0x80000000, // PA_SC_VPORT_SCISSOR_0_TL
0x40004000, // PA_SC_VPORT_SCISSOR_0_BR
0x80000000, // PA_SC_VPORT_SCISSOR_1_TL
0x40004000, // PA_SC_VPORT_SCISSOR_1_BR
0x80000000, // PA_SC_VPORT_SCISSOR_2_TL
0x40004000, // PA_SC_VPORT_SCISSOR_2_BR
0x80000000, // PA_SC_VPORT_SCISSOR_3_TL
0x40004000, // PA_SC_VPORT_SCISSOR_3_BR
0x80000000, // PA_SC_VPORT_SCISSOR_4_TL
0x40004000, // PA_SC_VPORT_SCISSOR_4_BR
0x80000000, // PA_SC_VPORT_SCISSOR_5_TL
0x40004000, // PA_SC_VPORT_SCISSOR_5_BR
0x80000000, // PA_SC_VPORT_SCISSOR_6_TL
0x40004000, // PA_SC_VPORT_SCISSOR_6_BR
0x80000000, // PA_SC_VPORT_SCISSOR_7_TL
0x40004000, // PA_SC_VPORT_SCISSOR_7_BR
0x80000000, // PA_SC_VPORT_SCISSOR_8_TL
0x40004000, // PA_SC_VPORT_SCISSOR_8_BR
0x80000000, // PA_SC_VPORT_SCISSOR_9_TL
0x40004000, // PA_SC_VPORT_SCISSOR_9_BR
0x80000000, // PA_SC_VPORT_SCISSOR_10_TL
0x40004000, // PA_SC_VPORT_SCISSOR_10_BR
0x80000000, // PA_SC_VPORT_SCISSOR_11_TL
0x40004000, // PA_SC_VPORT_SCISSOR_11_BR
0x80000000, // PA_SC_VPORT_SCISSOR_12_TL
0x40004000, // PA_SC_VPORT_SCISSOR_12_BR
0x80000000, // PA_SC_VPORT_SCISSOR_13_TL
0x40004000, // PA_SC_VPORT_SCISSOR_13_BR
0x80000000, // PA_SC_VPORT_SCISSOR_14_TL
0x40004000, // PA_SC_VPORT_SCISSOR_14_BR
0x80000000, // PA_SC_VPORT_SCISSOR_15_TL
0x40004000, // PA_SC_VPORT_SCISSOR_15_BR
0x00000000, // PA_SC_VPORT_ZMIN_0
0x3f800000, // PA_SC_VPORT_ZMAX_0
0x00000000, // PA_SC_VPORT_ZMIN_1
0x3f800000, // PA_SC_VPORT_ZMAX_1
0x00000000, // PA_SC_VPORT_ZMIN_2
0x3f800000, // PA_SC_VPORT_ZMAX_2
0x00000000, // PA_SC_VPORT_ZMIN_3
0x3f800000, // PA_SC_VPORT_ZMAX_3
0x00000000, // PA_SC_VPORT_ZMIN_4
0x3f800000, // PA_SC_VPORT_ZMAX_4
0x00000000, // PA_SC_VPORT_ZMIN_5
0x3f800000, // PA_SC_VPORT_ZMAX_5
0x00000000, // PA_SC_VPORT_ZMIN_6
0x3f800000, // PA_SC_VPORT_ZMAX_6
0x00000000, // PA_SC_VPORT_ZMIN_7
0x3f800000, // PA_SC_VPORT_ZMAX_7
0x00000000, // PA_SC_VPORT_ZMIN_8
0x3f800000, // PA_SC_VPORT_ZMAX_8
0x00000000, // PA_SC_VPORT_ZMIN_9
0x3f800000, // PA_SC_VPORT_ZMAX_9
0x00000000, // PA_SC_VPORT_ZMIN_10
0x3f800000, // PA_SC_VPORT_ZMAX_10
0x00000000, // PA_SC_VPORT_ZMIN_11
0x3f800000, // PA_SC_VPORT_ZMAX_11
0x00000000, // PA_SC_VPORT_ZMIN_12
0x3f800000, // PA_SC_VPORT_ZMAX_12
0x00000000, // PA_SC_VPORT_ZMIN_13
0x3f800000, // PA_SC_VPORT_ZMAX_13
0x00000000, // PA_SC_VPORT_ZMIN_14
0x3f800000, // PA_SC_VPORT_ZMAX_14
0x00000000, // PA_SC_VPORT_ZMIN_15
0x3f800000, // PA_SC_VPORT_ZMAX_15
0x00000000, // PA_SC_RASTER_CONFIG
0x00000000, // PA_SC_RASTER_CONFIG_1
0x00000000, // PA_SC_SCREEN_EXTENT_CONTROL
};
static const unsigned int gfx10_SECT_CONTEXT_def_2[] = {
0x00000000, // CP_PERFMON_CNTX_CNTL
0x00000000, // CP_RINGID
0x00000000, // CP_VMID
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // PA_SC_RIGHT_VERT_GRID
0x00000000, // PA_SC_LEFT_VERT_GRID
0x00000000, // PA_SC_HORIZ_GRID
0x00000000, // HOLE
0x00000000, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0xffffffff, // VGT_MAX_VTX_INDX
0x00000000, // VGT_MIN_VTX_INDX
0x00000000, // VGT_INDX_OFFSET
0x00000000, // VGT_MULTI_PRIM_IB_RESET_INDX
0x00550055, // CB_RMI_GL2_CACHE_CONTROL
0x00000000, // CB_BLEND_RED
0x00000000, // CB_BLEND_GREEN
0x00000000, // CB_BLEND_BLUE
0x00000000, // CB_BLEND_ALPHA
0x00000000, // CB_DCC_CONTROL
0x00000000, // CB_COVERAGE_OUT_CONTROL
0x00000000, // DB_STENCIL_CONTROL
0x01000000, // DB_STENCILREFMASK
0x01000000, // DB_STENCILREFMASK_BF
0, // HOLE
0x00000000, // PA_CL_VPORT_XSCALE
0x00000000, // PA_CL_VPORT_XOFFSET
0x00000000, // PA_CL_VPORT_YSCALE
0x00000000, // PA_CL_VPORT_YOFFSET
0x00000000, // PA_CL_VPORT_ZSCALE
0x00000000, // PA_CL_VPORT_ZOFFSET
0x00000000, // PA_CL_VPORT_XSCALE_1
0x00000000, // PA_CL_VPORT_XOFFSET_1
0x00000000, // PA_CL_VPORT_YSCALE_1
0x00000000, // PA_CL_VPORT_YOFFSET_1
0x00000000, // PA_CL_VPORT_ZSCALE_1
0x00000000, // PA_CL_VPORT_ZOFFSET_1
0x00000000, // PA_CL_VPORT_XSCALE_2
0x00000000, // PA_CL_VPORT_XOFFSET_2
0x00000000, // PA_CL_VPORT_YSCALE_2
0x00000000, // PA_CL_VPORT_YOFFSET_2
0x00000000, // PA_CL_VPORT_ZSCALE_2
0x00000000, // PA_CL_VPORT_ZOFFSET_2
0x00000000, // PA_CL_VPORT_XSCALE_3
0x00000000, // PA_CL_VPORT_XOFFSET_3
0x00000000, // PA_CL_VPORT_YSCALE_3
0x00000000, // PA_CL_VPORT_YOFFSET_3
0x00000000, // PA_CL_VPORT_ZSCALE_3
0x00000000, // PA_CL_VPORT_ZOFFSET_3
0x00000000, // PA_CL_VPORT_XSCALE_4
0x00000000, // PA_CL_VPORT_XOFFSET_4
0x00000000, // PA_CL_VPORT_YSCALE_4
0x00000000, // PA_CL_VPORT_YOFFSET_4
0x00000000, // PA_CL_VPORT_ZSCALE_4
0x00000000, // PA_CL_VPORT_ZOFFSET_4
0x00000000, // PA_CL_VPORT_XSCALE_5
0x00000000, // PA_CL_VPORT_XOFFSET_5
0x00000000, // PA_CL_VPORT_YSCALE_5
0x00000000, // PA_CL_VPORT_YOFFSET_5
0x00000000, // PA_CL_VPORT_ZSCALE_5
0x00000000, // PA_CL_VPORT_ZOFFSET_5
0x00000000, // PA_CL_VPORT_XSCALE_6
0x00000000, // PA_CL_VPORT_XOFFSET_6
0x00000000, // PA_CL_VPORT_YSCALE_6
0x00000000, // PA_CL_VPORT_YOFFSET_6
0x00000000, // PA_CL_VPORT_ZSCALE_6
0x00000000, // PA_CL_VPORT_ZOFFSET_6
0x00000000, // PA_CL_VPORT_XSCALE_7
0x00000000, // PA_CL_VPORT_XOFFSET_7
0x00000000, // PA_CL_VPORT_YSCALE_7
0x00000000, // PA_CL_VPORT_YOFFSET_7
0x00000000, // PA_CL_VPORT_ZSCALE_7
0x00000000, // PA_CL_VPORT_ZOFFSET_7
0x00000000, // PA_CL_VPORT_XSCALE_8
0x00000000, // PA_CL_VPORT_XOFFSET_8
0x00000000, // PA_CL_VPORT_YSCALE_8
0x00000000, // PA_CL_VPORT_YOFFSET_8
0x00000000, // PA_CL_VPORT_ZSCALE_8
0x00000000, // PA_CL_VPORT_ZOFFSET_8
0x00000000, // PA_CL_VPORT_XSCALE_9
0x00000000, // PA_CL_VPORT_XOFFSET_9
0x00000000, // PA_CL_VPORT_YSCALE_9
0x00000000, // PA_CL_VPORT_YOFFSET_9
0x00000000, // PA_CL_VPORT_ZSCALE_9
0x00000000, // PA_CL_VPORT_ZOFFSET_9
0x00000000, // PA_CL_VPORT_XSCALE_10
0x00000000, // PA_CL_VPORT_XOFFSET_10
0x00000000, // PA_CL_VPORT_YSCALE_10
0x00000000, // PA_CL_VPORT_YOFFSET_10
0x00000000, // PA_CL_VPORT_ZSCALE_10
0x00000000, // PA_CL_VPORT_ZOFFSET_10
0x00000000, // PA_CL_VPORT_XSCALE_11
0x00000000, // PA_CL_VPORT_XOFFSET_11
0x00000000, // PA_CL_VPORT_YSCALE_11
0x00000000, // PA_CL_VPORT_YOFFSET_11
0x00000000, // PA_CL_VPORT_ZSCALE_11
0x00000000, // PA_CL_VPORT_ZOFFSET_11
0x00000000, // PA_CL_VPORT_XSCALE_12
0x00000000, // PA_CL_VPORT_XOFFSET_12
0x00000000, // PA_CL_VPORT_YSCALE_12
0x00000000, // PA_CL_VPORT_YOFFSET_12
0x00000000, // PA_CL_VPORT_ZSCALE_12
0x00000000, // PA_CL_VPORT_ZOFFSET_12
0x00000000, // PA_CL_VPORT_XSCALE_13
0x00000000, // PA_CL_VPORT_XOFFSET_13
0x00000000, // PA_CL_VPORT_YSCALE_13
0x00000000, // PA_CL_VPORT_YOFFSET_13
0x00000000, // PA_CL_VPORT_ZSCALE_13
0x00000000, // PA_CL_VPORT_ZOFFSET_13
0x00000000, // PA_CL_VPORT_XSCALE_14
0x00000000, // PA_CL_VPORT_XOFFSET_14
0x00000000, // PA_CL_VPORT_YSCALE_14
0x00000000, // PA_CL_VPORT_YOFFSET_14
0x00000000, // PA_CL_VPORT_ZSCALE_14
0x00000000, // PA_CL_VPORT_ZOFFSET_14
0x00000000, // PA_CL_VPORT_XSCALE_15
0x00000000, // PA_CL_VPORT_XOFFSET_15
0x00000000, // PA_CL_VPORT_YSCALE_15
0x00000000, // PA_CL_VPORT_YOFFSET_15
0x00000000, // PA_CL_VPORT_ZSCALE_15
0x00000000, // PA_CL_VPORT_ZOFFSET_15
0x00000000, // PA_CL_UCP_0_X
0x00000000, // PA_CL_UCP_0_Y
0x00000000, // PA_CL_UCP_0_Z
0x00000000, // PA_CL_UCP_0_W
0x00000000, // PA_CL_UCP_1_X
0x00000000, // PA_CL_UCP_1_Y
0x00000000, // PA_CL_UCP_1_Z
0x00000000, // PA_CL_UCP_1_W
0x00000000, // PA_CL_UCP_2_X
0x00000000, // PA_CL_UCP_2_Y
0x00000000, // PA_CL_UCP_2_Z
0x00000000, // PA_CL_UCP_2_W
0x00000000, // PA_CL_UCP_3_X
0x00000000, // PA_CL_UCP_3_Y
0x00000000, // PA_CL_UCP_3_Z
0x00000000, // PA_CL_UCP_3_W
0x00000000, // PA_CL_UCP_4_X
0x00000000, // PA_CL_UCP_4_Y
0x00000000, // PA_CL_UCP_4_Z
0x00000000, // PA_CL_UCP_4_W
0x00000000, // PA_CL_UCP_5_X
0x00000000, // PA_CL_UCP_5_Y
0x00000000, // PA_CL_UCP_5_Z
0x00000000, // PA_CL_UCP_5_W
0x00000000, // PA_CL_PROG_NEAR_CLIP_Z
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // SPI_PS_INPUT_CNTL_0
0x00000000, // SPI_PS_INPUT_CNTL_1
0x00000000, // SPI_PS_INPUT_CNTL_2
0x00000000, // SPI_PS_INPUT_CNTL_3
0x00000000, // SPI_PS_INPUT_CNTL_4
0x00000000, // SPI_PS_INPUT_CNTL_5
0x00000000, // SPI_PS_INPUT_CNTL_6
0x00000000, // SPI_PS_INPUT_CNTL_7
0x00000000, // SPI_PS_INPUT_CNTL_8
0x00000000, // SPI_PS_INPUT_CNTL_9
0x00000000, // SPI_PS_INPUT_CNTL_10
0x00000000, // SPI_PS_INPUT_CNTL_11
0x00000000, // SPI_PS_INPUT_CNTL_12
0x00000000, // SPI_PS_INPUT_CNTL_13
0x00000000, // SPI_PS_INPUT_CNTL_14
0x00000000, // SPI_PS_INPUT_CNTL_15
0x00000000, // SPI_PS_INPUT_CNTL_16
0x00000000, // SPI_PS_INPUT_CNTL_17
0x00000000, // SPI_PS_INPUT_CNTL_18
0x00000000, // SPI_PS_INPUT_CNTL_19
0x00000000, // SPI_PS_INPUT_CNTL_20
0x00000000, // SPI_PS_INPUT_CNTL_21
0x00000000, // SPI_PS_INPUT_CNTL_22
0x00000000, // SPI_PS_INPUT_CNTL_23
0x00000000, // SPI_PS_INPUT_CNTL_24
0x00000000, // SPI_PS_INPUT_CNTL_25
0x00000000, // SPI_PS_INPUT_CNTL_26
0x00000000, // SPI_PS_INPUT_CNTL_27
0x00000000, // SPI_PS_INPUT_CNTL_28
0x00000000, // SPI_PS_INPUT_CNTL_29
0x00000000, // SPI_PS_INPUT_CNTL_30
0x00000000, // SPI_PS_INPUT_CNTL_31
0x00000000, // SPI_VS_OUT_CONFIG
0, // HOLE
0x00000000, // SPI_PS_INPUT_ENA
0x00000000, // SPI_PS_INPUT_ADDR
0x00000000, // SPI_INTERP_CONTROL_0
0x00000002, // SPI_PS_IN_CONTROL
0, // HOLE
0x00000000, // SPI_BARYC_CNTL
0, // HOLE
0x00000000, // SPI_TMPRING_SIZE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // SPI_SHADER_IDX_FORMAT
0x00000000, // SPI_SHADER_POS_FORMAT
0x00000000, // SPI_SHADER_Z_FORMAT
0x00000000, // SPI_SHADER_COL_FORMAT
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // SX_PS_DOWNCONVERT
0x00000000, // SX_BLEND_OPT_EPSILON
0x00000000, // SX_BLEND_OPT_CONTROL
0x00000000, // SX_MRT0_BLEND_OPT
0x00000000, // SX_MRT1_BLEND_OPT
0x00000000, // SX_MRT2_BLEND_OPT
0x00000000, // SX_MRT3_BLEND_OPT
0x00000000, // SX_MRT4_BLEND_OPT
0x00000000, // SX_MRT5_BLEND_OPT
0x00000000, // SX_MRT6_BLEND_OPT
0x00000000, // SX_MRT7_BLEND_OPT
0x00000000, // CB_BLEND0_CONTROL
0x00000000, // CB_BLEND1_CONTROL
0x00000000, // CB_BLEND2_CONTROL
0x00000000, // CB_BLEND3_CONTROL
0x00000000, // CB_BLEND4_CONTROL
0x00000000, // CB_BLEND5_CONTROL
0x00000000, // CB_BLEND6_CONTROL
0x00000000, // CB_BLEND7_CONTROL
};
static const unsigned int gfx10_SECT_CONTEXT_def_3[] = {
0x00000000, // PA_CL_POINT_X_RAD
0x00000000, // PA_CL_POINT_Y_RAD
0x00000000, // PA_CL_POINT_SIZE
0x00000000, // PA_CL_POINT_CULL_RAD
};
static const unsigned int gfx10_SECT_CONTEXT_def_4[] = {
0x00000000, // VGT_GS_MAX_PRIMS_PER_SUBGROUP
0x00000000, // DB_DEPTH_CONTROL
0x00000000, // DB_EQAA
0x00000000, // CB_COLOR_CONTROL
0x00000000, // DB_SHADER_CONTROL
0x00090000, // PA_CL_CLIP_CNTL
0x00000004, // PA_SU_SC_MODE_CNTL
0x00000000, // PA_CL_VTE_CNTL
0x00000000, // PA_CL_VS_OUT_CNTL
0x00000000, // PA_CL_NANINF_CNTL
0x00000000, // PA_SU_LINE_STIPPLE_CNTL
0x00000000, // PA_SU_LINE_STIPPLE_SCALE
0x00000000, // PA_SU_PRIM_FILTER_CNTL
0x00000000, // PA_SU_SMALL_PRIM_FILTER_CNTL
0x00000000, // PA_CL_OBJPRIM_ID_CNTL
0x00000000, // PA_CL_NGG_CNTL
0x00000000, // PA_SU_OVER_RASTERIZATION_CNTL
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // PA_SU_POINT_SIZE
0x00000000, // PA_SU_POINT_MINMAX
0x00000000, // PA_SU_LINE_CNTL
0x00000000, // PA_SC_LINE_STIPPLE
0x00000000, // VGT_OUTPUT_PATH_CNTL
0x00000000, // VGT_HOS_CNTL
0x00000000, // VGT_HOS_MAX_TESS_LEVEL
0x00000000, // VGT_HOS_MIN_TESS_LEVEL
0x00000000, // VGT_HOS_REUSE_DEPTH
0x00000000, // VGT_GROUP_PRIM_TYPE
0x00000000, // VGT_GROUP_FIRST_DECR
0x00000000, // VGT_GROUP_DECR
0x00000000, // VGT_GROUP_VECT_0_CNTL
0x00000000, // VGT_GROUP_VECT_1_CNTL
0x00000000, // VGT_GROUP_VECT_0_FMT_CNTL
0x00000000, // VGT_GROUP_VECT_1_FMT_CNTL
0x00000000, // VGT_GS_MODE
0x00000000, // VGT_GS_ONCHIP_CNTL
0x00000000, // PA_SC_MODE_CNTL_0
0x00000000, // PA_SC_MODE_CNTL_1
0x00000000, // VGT_ENHANCE
0x00000100, // VGT_GS_PER_ES
0x00000080, // VGT_ES_PER_GS
0x00000002, // VGT_GS_PER_VS
0x00000000, // VGT_GSVS_RING_OFFSET_1
0x00000000, // VGT_GSVS_RING_OFFSET_2
0x00000000, // VGT_GSVS_RING_OFFSET_3
0x00000000, // VGT_GS_OUT_PRIM_TYPE
0x00000000, // IA_ENHANCE
};
static const unsigned int gfx10_SECT_CONTEXT_def_5[] = {
0x00000000, // WD_ENHANCE
0x00000000, // VGT_PRIMITIVEID_EN
};
static const unsigned int gfx10_SECT_CONTEXT_def_6[] = {
0x00000000, // VGT_PRIMITIVEID_RESET
};
static const unsigned int gfx10_SECT_CONTEXT_def_7[] = {
0x00000000, // VGT_MULTI_PRIM_IB_RESET_EN
0x00000000, // VGT_DRAW_PAYLOAD_CNTL
0x00000000, // HOLE
0x00000000, // VGT_INSTANCE_STEP_RATE_0
0x00000000, // VGT_INSTANCE_STEP_RATE_1
0x000000ff, // IA_MULTI_VGT_PARAM
0x00000000, // VGT_ESGS_RING_ITEMSIZE
0x00000000, // VGT_GSVS_RING_ITEMSIZE
0x00000000, // VGT_REUSE_OFF
0x00000000, // VGT_VTX_CNT_EN
0x00000000, // DB_HTILE_SURFACE
0x00000000, // DB_SRESULTS_COMPARE_STATE0
0x00000000, // DB_SRESULTS_COMPARE_STATE1
0x00000000, // DB_PRELOAD_CONTROL
0, // HOLE
0x00000000, // VGT_STRMOUT_BUFFER_SIZE_0
0x00000000, // VGT_STRMOUT_VTX_STRIDE_0
0, // HOLE
0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_0
0x00000000, // VGT_STRMOUT_BUFFER_SIZE_1
0x00000000, // VGT_STRMOUT_VTX_STRIDE_1
0, // HOLE
0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_1
0x00000000, // VGT_STRMOUT_BUFFER_SIZE_2
0x00000000, // VGT_STRMOUT_VTX_STRIDE_2
0, // HOLE
0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_2
0x00000000, // VGT_STRMOUT_BUFFER_SIZE_3
0x00000000, // VGT_STRMOUT_VTX_STRIDE_3
0, // HOLE
0x00000000, // VGT_STRMOUT_BUFFER_OFFSET_3
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_OFFSET
0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_BUFFER_FILLED_SIZE
0x00000000, // VGT_STRMOUT_DRAW_OPAQUE_VERTEX_STRIDE
0, // HOLE
0x00000000, // VGT_GS_MAX_VERT_OUT
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0, // HOLE
0x00000000, // VGT_TESS_DISTRIBUTION
0x00000000, // VGT_SHADER_STAGES_EN
0x00000000, // VGT_LS_HS_CONFIG
0x00000000, // VGT_GS_VERT_ITEMSIZE
0x00000000, // VGT_GS_VERT_ITEMSIZE_1
0x00000000, // VGT_GS_VERT_ITEMSIZE_2
0x00000000, // VGT_GS_VERT_ITEMSIZE_3
0x00000000, // VGT_TF_PARAM
0x00000000, // DB_ALPHA_TO_MASK
0x00000000, // VGT_DISPATCH_DRAW_INDEX
0x00000000, // PA_SU_POLY_OFFSET_DB_FMT_CNTL
0x00000000, // PA_SU_POLY_OFFSET_CLAMP
0x00000000, // PA_SU_POLY_OFFSET_FRONT_SCALE
0x00000000, // PA_SU_POLY_OFFSET_FRONT_OFFSET
0x00000000, // PA_SU_POLY_OFFSET_BACK_SCALE
0x00000000, // PA_SU_POLY_OFFSET_BACK_OFFSET
0x00000000, // VGT_GS_INSTANCE_CNT
0x00000000, // VGT_STRMOUT_CONFIG
0x00000000, // VGT_STRMOUT_BUFFER_CONFIG
};
static const unsigned int gfx10_SECT_CONTEXT_def_8[] = {
0x00000000, // PA_SC_CENTROID_PRIORITY_0
0x00000000, // PA_SC_CENTROID_PRIORITY_1
0x00001000, // PA_SC_LINE_CNTL
0x00000000, // PA_SC_AA_CONFIG
0x00000005, // PA_SU_VTX_CNTL
0x3f800000, // PA_CL_GB_VERT_CLIP_ADJ
0x3f800000, // PA_CL_GB_VERT_DISC_ADJ
0x3f800000, // PA_CL_GB_HORZ_CLIP_ADJ
0x3f800000, // PA_CL_GB_HORZ_DISC_ADJ
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2
0x00000000, // PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3
0xffffffff, // PA_SC_AA_MASK_X0Y0_X1Y0
0xffffffff, // PA_SC_AA_MASK_X0Y1_X1Y1
0x00000000, // PA_SC_SHADER_CONTROL
0x00000003, // PA_SC_BINNER_CNTL_0
0x00000000, // PA_SC_BINNER_CNTL_1
0x00100000, // PA_SC_CONSERVATIVE_RASTERIZATION_CNTL
0x00000000, // PA_SC_NGG_MODE_CNTL
0, // HOLE
0x0000001e, // VGT_VERTEX_REUSE_BLOCK_CNTL
0x00000020, // VGT_OUT_DEALLOC_CNTL
0x00000000, // CB_COLOR0_BASE
0x00000000, // CB_COLOR0_PITCH
0x00000000, // CB_COLOR0_SLICE
0x00000000, // CB_COLOR0_VIEW
0x00000000, // CB_COLOR0_INFO
0x00000000, // CB_COLOR0_ATTRIB
0x00000000, // CB_COLOR0_DCC_CONTROL
0x00000000, // CB_COLOR0_CMASK
0x00000000, // CB_COLOR0_CMASK_SLICE
0x00000000, // CB_COLOR0_FMASK
0x00000000, // CB_COLOR0_FMASK_SLICE
0x00000000, // CB_COLOR0_CLEAR_WORD0
0x00000000, // CB_COLOR0_CLEAR_WORD1
0x00000000, // CB_COLOR0_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR1_BASE
0x00000000, // CB_COLOR1_PITCH
0x00000000, // CB_COLOR1_SLICE
0x00000000, // CB_COLOR1_VIEW
0x00000000, // CB_COLOR1_INFO
0x00000000, // CB_COLOR1_ATTRIB
0x00000000, // CB_COLOR1_DCC_CONTROL
0x00000000, // CB_COLOR1_CMASK
0x00000000, // CB_COLOR1_CMASK_SLICE
0x00000000, // CB_COLOR1_FMASK
0x00000000, // CB_COLOR1_FMASK_SLICE
0x00000000, // CB_COLOR1_CLEAR_WORD0
0x00000000, // CB_COLOR1_CLEAR_WORD1
0x00000000, // CB_COLOR1_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR2_BASE
0x00000000, // CB_COLOR2_PITCH
0x00000000, // CB_COLOR2_SLICE
0x00000000, // CB_COLOR2_VIEW
0x00000000, // CB_COLOR2_INFO
0x00000000, // CB_COLOR2_ATTRIB
0x00000000, // CB_COLOR2_DCC_CONTROL
0x00000000, // CB_COLOR2_CMASK
0x00000000, // CB_COLOR2_CMASK_SLICE
0x00000000, // CB_COLOR2_FMASK
0x00000000, // CB_COLOR2_FMASK_SLICE
0x00000000, // CB_COLOR2_CLEAR_WORD0
0x00000000, // CB_COLOR2_CLEAR_WORD1
0x00000000, // CB_COLOR2_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR3_BASE
0x00000000, // CB_COLOR3_PITCH
0x00000000, // CB_COLOR3_SLICE
0x00000000, // CB_COLOR3_VIEW
0x00000000, // CB_COLOR3_INFO
0x00000000, // CB_COLOR3_ATTRIB
0x00000000, // CB_COLOR3_DCC_CONTROL
0x00000000, // CB_COLOR3_CMASK
0x00000000, // CB_COLOR3_CMASK_SLICE
0x00000000, // CB_COLOR3_FMASK
0x00000000, // CB_COLOR3_FMASK_SLICE
0x00000000, // CB_COLOR3_CLEAR_WORD0
0x00000000, // CB_COLOR3_CLEAR_WORD1
0x00000000, // CB_COLOR3_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR4_BASE
0x00000000, // CB_COLOR4_PITCH
0x00000000, // CB_COLOR4_SLICE
0x00000000, // CB_COLOR4_VIEW
0x00000000, // CB_COLOR4_INFO
0x00000000, // CB_COLOR4_ATTRIB
0x00000000, // CB_COLOR4_DCC_CONTROL
0x00000000, // CB_COLOR4_CMASK
0x00000000, // CB_COLOR4_CMASK_SLICE
0x00000000, // CB_COLOR4_FMASK
0x00000000, // CB_COLOR4_FMASK_SLICE
0x00000000, // CB_COLOR4_CLEAR_WORD0
0x00000000, // CB_COLOR4_CLEAR_WORD1
0x00000000, // CB_COLOR4_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR5_BASE
0x00000000, // CB_COLOR5_PITCH
0x00000000, // CB_COLOR5_SLICE
0x00000000, // CB_COLOR5_VIEW
0x00000000, // CB_COLOR5_INFO
0x00000000, // CB_COLOR5_ATTRIB
0x00000000, // CB_COLOR5_DCC_CONTROL
0x00000000, // CB_COLOR5_CMASK
0x00000000, // CB_COLOR5_CMASK_SLICE
0x00000000, // CB_COLOR5_FMASK
0x00000000, // CB_COLOR5_FMASK_SLICE
0x00000000, // CB_COLOR5_CLEAR_WORD0
0x00000000, // CB_COLOR5_CLEAR_WORD1
0x00000000, // CB_COLOR5_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR6_BASE
0x00000000, // CB_COLOR6_PITCH
0x00000000, // CB_COLOR6_SLICE
0x00000000, // CB_COLOR6_VIEW
0x00000000, // CB_COLOR6_INFO
0x00000000, // CB_COLOR6_ATTRIB
0x00000000, // CB_COLOR6_DCC_CONTROL
0x00000000, // CB_COLOR6_CMASK
0x00000000, // CB_COLOR6_CMASK_SLICE
0x00000000, // CB_COLOR6_FMASK
0x00000000, // CB_COLOR6_FMASK_SLICE
0x00000000, // CB_COLOR6_CLEAR_WORD0
0x00000000, // CB_COLOR6_CLEAR_WORD1
0x00000000, // CB_COLOR6_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR7_BASE
0x00000000, // CB_COLOR7_PITCH
0x00000000, // CB_COLOR7_SLICE
0x00000000, // CB_COLOR7_VIEW
0x00000000, // CB_COLOR7_INFO
0x00000000, // CB_COLOR7_ATTRIB
0x00000000, // CB_COLOR7_DCC_CONTROL
0x00000000, // CB_COLOR7_CMASK
0x00000000, // CB_COLOR7_CMASK_SLICE
0x00000000, // CB_COLOR7_FMASK
0x00000000, // CB_COLOR7_FMASK_SLICE
0x00000000, // CB_COLOR7_CLEAR_WORD0
0x00000000, // CB_COLOR7_CLEAR_WORD1
0x00000000, // CB_COLOR7_DCC_BASE
0, // HOLE
0x00000000, // CB_COLOR0_BASE_EXT
0x00000000, // CB_COLOR1_BASE_EXT
0x00000000, // CB_COLOR2_BASE_EXT
0x00000000, // CB_COLOR3_BASE_EXT
0x00000000, // CB_COLOR4_BASE_EXT
0x00000000, // CB_COLOR5_BASE_EXT
0x00000000, // CB_COLOR6_BASE_EXT
0x00000000, // CB_COLOR7_BASE_EXT
0x00000000, // CB_COLOR0_CMASK_BASE_EXT
0x00000000, // CB_COLOR1_CMASK_BASE_EXT
0x00000000, // CB_COLOR2_CMASK_BASE_EXT
0x00000000, // CB_COLOR3_CMASK_BASE_EXT
0x00000000, // CB_COLOR4_CMASK_BASE_EXT
0x00000000, // CB_COLOR5_CMASK_BASE_EXT
0x00000000, // CB_COLOR6_CMASK_BASE_EXT
0x00000000, // CB_COLOR7_CMASK_BASE_EXT
0x00000000, // CB_COLOR0_FMASK_BASE_EXT
0x00000000, // CB_COLOR1_FMASK_BASE_EXT
0x00000000, // CB_COLOR2_FMASK_BASE_EXT
0x00000000, // CB_COLOR3_FMASK_BASE_EXT
0x00000000, // CB_COLOR4_FMASK_BASE_EXT
0x00000000, // CB_COLOR5_FMASK_BASE_EXT
0x00000000, // CB_COLOR6_FMASK_BASE_EXT
0x00000000, // CB_COLOR7_FMASK_BASE_EXT
0x00000000, // CB_COLOR0_DCC_BASE_EXT
0x00000000, // CB_COLOR1_DCC_BASE_EXT
0x00000000, // CB_COLOR2_DCC_BASE_EXT
0x00000000, // CB_COLOR3_DCC_BASE_EXT
0x00000000, // CB_COLOR4_DCC_BASE_EXT
0x00000000, // CB_COLOR5_DCC_BASE_EXT
0x00000000, // CB_COLOR6_DCC_BASE_EXT
0x00000000, // CB_COLOR7_DCC_BASE_EXT
0x00000000, // CB_COLOR0_ATTRIB2
0x00000000, // CB_COLOR1_ATTRIB2
0x00000000, // CB_COLOR2_ATTRIB2
0x00000000, // CB_COLOR3_ATTRIB2
0x00000000, // CB_COLOR4_ATTRIB2
0x00000000, // CB_COLOR5_ATTRIB2
0x00000000, // CB_COLOR6_ATTRIB2
0x00000000, // CB_COLOR7_ATTRIB2
0x00000000, // CB_COLOR0_ATTRIB3
0x00000000, // CB_COLOR1_ATTRIB3
0x00000000, // CB_COLOR2_ATTRIB3
0x00000000, // CB_COLOR3_ATTRIB3
0x00000000, // CB_COLOR4_ATTRIB3
0x00000000, // CB_COLOR5_ATTRIB3
0x00000000, // CB_COLOR6_ATTRIB3
0x00000000, // CB_COLOR7_ATTRIB3
};
static const struct cs_extent_def gfx10_SECT_CONTEXT_defs[] = {
{gfx10_SECT_CONTEXT_def_1, 0x0000a000, 215 },
{gfx10_SECT_CONTEXT_def_2, 0x0000a0d8, 272 },
{gfx10_SECT_CONTEXT_def_3, 0x0000a1f5, 4 },
{gfx10_SECT_CONTEXT_def_4, 0x0000a1ff, 158 },
{gfx10_SECT_CONTEXT_def_5, 0x0000a2a0, 2 },
{gfx10_SECT_CONTEXT_def_6, 0x0000a2a3, 1 },
{gfx10_SECT_CONTEXT_def_7, 0x0000a2a5, 66 },
{gfx10_SECT_CONTEXT_def_8, 0x0000a2f5, 203 },
{ 0, 0, 0 }
};
static const struct cs_section_def gfx10_cs_data[] = {
{ gfx10_SECT_CONTEXT_defs, SECT_CONTEXT },
{ 0, SECT_NONE }
};

1
drivers/gpu/drm/amd/amdgpu/dce_virtual.c
Просмотреть файл

@ -457,6 +457,7 @@ static int dce_virtual_hw_init(void *handle)
case CHIP_VEGA10:
case CHIP_VEGA12:
case CHIP_VEGA20:
case CHIP_NAVI10:
break;
default:
DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type);

4
drivers/gpu/drm/amd/amdgpu/df_v1_7.c
Просмотреть файл

@ -29,7 +29,7 @@
static u32 df_v1_7_channel_number[] = {1, 2, 0, 4, 0, 8, 0, 16, 2};
static void df_v1_7_init (struct amdgpu_device *adev)
static void df_v1_7_sw_init(struct amdgpu_device *adev)
{
}
@ -110,7 +110,7 @@ static void df_v1_7_enable_ecc_force_par_wr_rmw(struct amdgpu_device *adev,
}
const struct amdgpu_df_funcs df_v1_7_funcs = {
.init = df_v1_7_init,
.sw_init = df_v1_7_sw_init,
.enable_broadcast_mode = df_v1_7_enable_broadcast_mode,
.get_fb_channel_number = df_v1_7_get_fb_channel_number,
.get_hbm_channel_number = df_v1_7_get_hbm_channel_number,

455
drivers/gpu/drm/amd/amdgpu/df_v3_6.c
Просмотреть файл

@ -30,8 +30,104 @@
static u32 df_v3_6_channel_number[] = {1, 2, 0, 4, 0, 8, 0,
16, 32, 0, 0, 0, 2, 4, 8};
static void df_v3_6_init(struct amdgpu_device *adev)
/* init df format attrs */
AMDGPU_PMU_ATTR(event, "config:0-7");
AMDGPU_PMU_ATTR(instance, "config:8-15");
AMDGPU_PMU_ATTR(umask, "config:16-23");
/* df format attributes */
static struct attribute *df_v3_6_format_attrs[] = {
&pmu_attr_event.attr,
&pmu_attr_instance.attr,
&pmu_attr_umask.attr,
NULL
};
/* df format attribute group */
static struct attribute_group df_v3_6_format_attr_group = {
.name = "format",
.attrs = df_v3_6_format_attrs,
};
/* df event attrs */
AMDGPU_PMU_ATTR(cake0_pcsout_txdata,
"event=0x7,instance=0x46,umask=0x2");
AMDGPU_PMU_ATTR(cake1_pcsout_txdata,
"event=0x7,instance=0x47,umask=0x2");
AMDGPU_PMU_ATTR(cake0_pcsout_txmeta,
"event=0x7,instance=0x46,umask=0x4");
AMDGPU_PMU_ATTR(cake1_pcsout_txmeta,
"event=0x7,instance=0x47,umask=0x4");
AMDGPU_PMU_ATTR(cake0_ftiinstat_reqalloc,
"event=0xb,instance=0x46,umask=0x4");
AMDGPU_PMU_ATTR(cake1_ftiinstat_reqalloc,
"event=0xb,instance=0x47,umask=0x4");
AMDGPU_PMU_ATTR(cake0_ftiinstat_rspalloc,
"event=0xb,instance=0x46,umask=0x8");
AMDGPU_PMU_ATTR(cake1_ftiinstat_rspalloc,
"event=0xb,instance=0x47,umask=0x8");
/* df event attributes */
static struct attribute *df_v3_6_event_attrs[] = {
&pmu_attr_cake0_pcsout_txdata.attr,
&pmu_attr_cake1_pcsout_txdata.attr,
&pmu_attr_cake0_pcsout_txmeta.attr,
&pmu_attr_cake1_pcsout_txmeta.attr,
&pmu_attr_cake0_ftiinstat_reqalloc.attr,
&pmu_attr_cake1_ftiinstat_reqalloc.attr,
&pmu_attr_cake0_ftiinstat_rspalloc.attr,
&pmu_attr_cake1_ftiinstat_rspalloc.attr,
NULL
};
/* df event attribute group */
static struct attribute_group df_v3_6_event_attr_group = {
.name = "events",
.attrs = df_v3_6_event_attrs
};
/* df event attr groups */
const struct attribute_group *df_v3_6_attr_groups[] = {
&df_v3_6_format_attr_group,
&df_v3_6_event_attr_group,
NULL
};
/* get the number of df counters available */
static ssize_t df_v3_6_get_df_cntr_avail(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct amdgpu_device *adev;
struct drm_device *ddev;
int i, count;
ddev = dev_get_drvdata(dev);
adev = ddev->dev_private;
count = 0;
for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) {
if (adev->df_perfmon_config_assign_mask[i] == 0)
count++;
}
return snprintf(buf, PAGE_SIZE, "%i\n", count);
}
/* device attr for available perfmon counters */
static DEVICE_ATTR(df_cntr_avail, S_IRUGO, df_v3_6_get_df_cntr_avail, NULL);
/* init perfmons */
static void df_v3_6_sw_init(struct amdgpu_device *adev)
{
int i, ret;
ret = device_create_file(adev->dev, &dev_attr_df_cntr_avail);
if (ret)
DRM_ERROR("failed to create file for available df counters\n");
for (i = 0; i < AMDGPU_MAX_DF_PERFMONS; i++)
adev->df_perfmon_config_assign_mask[i] = 0;
}
static void df_v3_6_enable_broadcast_mode(struct amdgpu_device *adev,
@ -105,28 +201,19 @@ static void df_v3_6_get_clockgating_state(struct amdgpu_device *adev,
*flags |= AMD_CG_SUPPORT_DF_MGCG;
}
/* hold counter assignment per gpu struct */
struct df_v3_6_event_mask {
struct amdgpu_device gpu;
uint64_t config_assign_mask[AMDGPU_DF_MAX_COUNTERS];
};
/* get assigned df perfmon ctr as int */
static void df_v3_6_pmc_config_2_cntr(struct amdgpu_device *adev,
uint64_t config,
int *counter)
static int df_v3_6_pmc_config_2_cntr(struct amdgpu_device *adev,
uint64_t config)
{
struct df_v3_6_event_mask *mask;
int i;
mask = container_of(adev, struct df_v3_6_event_mask, gpu);
for (i = 0; i < AMDGPU_DF_MAX_COUNTERS; i++) {
if ((config & 0x0FFFFFFUL) == mask->config_assign_mask[i]) {
*counter = i;
return;
}
for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) {
if ((config & 0x0FFFFFFUL) ==
adev->df_perfmon_config_assign_mask[i])
return i;
}
return -EINVAL;
}
/* get address based on counter assignment */
@ -136,10 +223,7 @@ static void df_v3_6_pmc_get_addr(struct amdgpu_device *adev,
uint32_t *lo_base_addr,
uint32_t *hi_base_addr)
{
int target_cntr = -1;
df_v3_6_pmc_config_2_cntr(adev, config, &target_cntr);
int target_cntr = df_v3_6_pmc_config_2_cntr(adev, config);
if (target_cntr < 0)
return;
@ -177,40 +261,38 @@ static void df_v3_6_pmc_get_read_settings(struct amdgpu_device *adev,
}
/* get control counter settings i.e. address and values to set */
static void df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev,
static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev,
uint64_t config,
uint32_t *lo_base_addr,
uint32_t *hi_base_addr,
uint32_t *lo_val,
uint32_t *hi_val)
{
uint32_t eventsel, instance, unitmask;
uint32_t es_5_0, es_13_0, es_13_6, es_13_12, es_11_8, es_7_0;
df_v3_6_pmc_get_addr(adev, config, 1, lo_base_addr, hi_base_addr);
if (lo_val == NULL || hi_val == NULL)
return;
if ((*lo_base_addr == 0) || (*hi_base_addr == 0)) {
DRM_ERROR("DF PMC addressing not retrieved! Lo: %x, Hi: %x",
DRM_ERROR("[DF PMC] addressing not retrieved! Lo: %x, Hi: %x",
*lo_base_addr, *hi_base_addr);
return;
return -ENXIO;
}
eventsel = GET_EVENT(config);
instance = GET_INSTANCE(config);
unitmask = GET_UNITMASK(config);
if (lo_val && hi_val) {
uint32_t eventsel, instance, unitmask;
uint32_t instance_10, instance_5432, instance_76;
es_5_0 = eventsel & 0x3FUL;
es_13_6 = instance;
es_13_0 = (es_13_6 << 6) + es_5_0;
es_13_12 = (es_13_0 & 0x03000UL) >> 12;
es_11_8 = (es_13_0 & 0x0F00UL) >> 8;
es_7_0 = es_13_0 & 0x0FFUL;
*lo_val = (es_7_0 & 0xFFUL) | ((unitmask & 0x0FUL) << 8);
*hi_val = (es_11_8 | ((es_13_12)<<(29)));
eventsel = DF_V3_6_GET_EVENT(config) & 0x3f;
unitmask = DF_V3_6_GET_UNITMASK(config) & 0xf;
instance = DF_V3_6_GET_INSTANCE(config);
instance_10 = instance & 0x3;
instance_5432 = (instance >> 2) & 0xf;
instance_76 = (instance >> 6) & 0x3;
*lo_val = (unitmask << 8) | (instance_10 << 6) | eventsel;
*hi_val = (instance_76 << 29) | instance_5432;
}
return 0;
}
/* assign df performance counters for read */
@ -218,26 +300,21 @@ static int df_v3_6_pmc_assign_cntr(struct amdgpu_device *adev,
uint64_t config,
int *is_assigned)
{
struct df_v3_6_event_mask *mask;
int i, target_cntr;
target_cntr = -1;
*is_assigned = 0;
df_v3_6_pmc_config_2_cntr(adev, config, &target_cntr);
target_cntr = df_v3_6_pmc_config_2_cntr(adev, config);
if (target_cntr >= 0) {
*is_assigned = 1;
return 0;
}
mask = container_of(adev, struct df_v3_6_event_mask, gpu);
for (i = 0; i < AMDGPU_DF_MAX_COUNTERS; i++) {
if (mask->config_assign_mask[i] == 0ULL) {
mask->config_assign_mask[i] = config & 0x0FFFFFFUL;
for (i = 0; i < DF_V3_6_MAX_COUNTERS; i++) {
if (adev->df_perfmon_config_assign_mask[i] == 0U) {
adev->df_perfmon_config_assign_mask[i] =
config & 0x0FFFFFFUL;
return 0;
}
}
@ -249,66 +326,17 @@ static int df_v3_6_pmc_assign_cntr(struct amdgpu_device *adev,
static void df_v3_6_pmc_release_cntr(struct amdgpu_device *adev,
uint64_t config)
{
struct df_v3_6_event_mask *mask;
int target_cntr;
target_cntr = -1;
df_v3_6_pmc_config_2_cntr(adev, config, &target_cntr);
mask = container_of(adev, struct df_v3_6_event_mask, gpu);
int target_cntr = df_v3_6_pmc_config_2_cntr(adev, config);
if (target_cntr >= 0)
mask->config_assign_mask[target_cntr] = 0ULL;
adev->df_perfmon_config_assign_mask[target_cntr] = 0ULL;
}
/*
* get xgmi link counters via programmable data fabric (df) counters (max 4)
* using cake tx event.
*
* @adev -> amdgpu device
* @instance-> currently cake has 2 links to poll on vega20
* @count -> counters to pass
*
*/
static void df_v3_6_get_xgmi_link_cntr(struct amdgpu_device *adev,
int instance,
uint64_t *count)
{
uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
uint64_t config;
config = GET_INSTANCE_CONFIG(instance);
df_v3_6_pmc_get_read_settings(adev, config, &lo_base_addr,
&hi_base_addr);
if ((lo_base_addr == 0) || (hi_base_addr == 0))
return;
lo_val = RREG32_PCIE(lo_base_addr);
hi_val = RREG32_PCIE(hi_base_addr);
*count = ((hi_val | 0ULL) << 32) | (lo_val | 0ULL);
}
/*
* reset xgmi link counters
*
* @adev -> amdgpu device
* @instance-> currently cake has 2 links to poll on vega20
*
*/
static void df_v3_6_reset_xgmi_link_cntr(struct amdgpu_device *adev,
int instance)
static void df_v3_6_reset_perfmon_cntr(struct amdgpu_device *adev,
uint64_t config)
{
uint32_t lo_base_addr, hi_base_addr;
uint64_t config;
config = 0ULL | (0x7ULL) | ((0x46ULL + instance) << 8) | (0x2 << 16);
df_v3_6_pmc_get_read_settings(adev, config, &lo_base_addr,
&hi_base_addr);
@ -320,157 +348,69 @@ static void df_v3_6_reset_xgmi_link_cntr(struct amdgpu_device *adev,
WREG32_PCIE(hi_base_addr, 0UL);
}
/*
* add xgmi link counters
*
* @adev -> amdgpu device
* @instance-> currently cake has 2 links to poll on vega20
*
*/
static int df_v3_6_add_xgmi_link_cntr(struct amdgpu_device *adev,
int instance)
static int df_v3_6_add_perfmon_cntr(struct amdgpu_device *adev,
uint64_t config)
{
uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
uint64_t config;
int ret, is_assigned;
if (instance < 0 || instance > 1)
return -EINVAL;
config = GET_INSTANCE_CONFIG(instance);
ret = df_v3_6_pmc_assign_cntr(adev, config, &is_assigned);
if (ret || is_assigned)
return ret;
df_v3_6_pmc_get_ctrl_settings(adev,
ret = df_v3_6_pmc_get_ctrl_settings(adev,
config,
&lo_base_addr,
&hi_base_addr,
&lo_val,
&hi_val);
if (ret)
return ret;
DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x:%08x",
config, lo_base_addr, hi_base_addr, lo_val, hi_val);
WREG32_PCIE(lo_base_addr, lo_val);
WREG32_PCIE(hi_base_addr, hi_val);
return ret;
}
/*
* start xgmi link counters
*
* @adev -> amdgpu device
* @instance-> currently cake has 2 links to poll on vega20
* @is_enable -> either resume or assign event via df perfmon
*
*/
static int df_v3_6_start_xgmi_link_cntr(struct amdgpu_device *adev,
int instance,
int is_enable)
{
uint32_t lo_base_addr, hi_base_addr, lo_val;
uint64_t config;
int ret;
if (instance < 0 || instance > 1)
return -EINVAL;
if (is_enable) {
ret = df_v3_6_add_xgmi_link_cntr(adev, instance);
if (ret)
return ret;
} else {
config = GET_INSTANCE_CONFIG(instance);
df_v3_6_pmc_get_ctrl_settings(adev,
config,
&lo_base_addr,
&hi_base_addr,
NULL,
NULL);
if (lo_base_addr == 0)
return -EINVAL;
lo_val = RREG32_PCIE(lo_base_addr);
WREG32_PCIE(lo_base_addr, lo_val | (1ULL << 22));
ret = 0;
}
return ret;
}
/*
* start xgmi link counters
*
* @adev -> amdgpu device
* @instance-> currently cake has 2 links to poll on vega20
* @is_enable -> either pause or unassign event via df perfmon
*
*/
static int df_v3_6_stop_xgmi_link_cntr(struct amdgpu_device *adev,
int instance,
int is_disable)
{
uint32_t lo_base_addr, hi_base_addr, lo_val;
uint64_t config;
config = GET_INSTANCE_CONFIG(instance);
if (is_disable) {
df_v3_6_reset_xgmi_link_cntr(adev, instance);
df_v3_6_pmc_release_cntr(adev, config);
} else {
df_v3_6_pmc_get_ctrl_settings(adev,
config,
&lo_base_addr,
&hi_base_addr,
NULL,
NULL);
if ((lo_base_addr == 0) || (hi_base_addr == 0))
return -EINVAL;
lo_val = RREG32_PCIE(lo_base_addr);
WREG32_PCIE(lo_base_addr, lo_val & ~(1ULL << 22));
}
return 0;
}
static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
int is_enable)
{
int xgmi_tx_link, ret = 0;
uint32_t lo_base_addr, hi_base_addr, lo_val;
int ret = 0;
switch (adev->asic_type) {
case CHIP_VEGA20:
xgmi_tx_link = IS_DF_XGMI_0_TX(config) ? 0
: (IS_DF_XGMI_1_TX(config) ? 1 : -1);
if (xgmi_tx_link >= 0)
ret = df_v3_6_start_xgmi_link_cntr(adev, xgmi_tx_link,
is_enable);
df_v3_6_reset_perfmon_cntr(adev, config);
if (ret)
return ret;
if (is_enable) {
ret = df_v3_6_add_perfmon_cntr(adev, config);
} else {
ret = df_v3_6_pmc_get_ctrl_settings(adev,
config,
&lo_base_addr,
&hi_base_addr,
NULL,
NULL);
if (ret)
return ret;
lo_val = RREG32_PCIE(lo_base_addr);
DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x",
config, lo_base_addr, hi_base_addr, lo_val);
WREG32_PCIE(lo_base_addr, lo_val | (1ULL << 22));
}
ret = 0;
break;
default:
break;
@ -482,23 +422,32 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config,
int is_disable)
{
int xgmi_tx_link, ret = 0;
uint32_t lo_base_addr, hi_base_addr, lo_val;
int ret = 0;
switch (adev->asic_type) {
case CHIP_VEGA20:
xgmi_tx_link = IS_DF_XGMI_0_TX(config) ? 0
: (IS_DF_XGMI_1_TX(config) ? 1 : -1);
ret = df_v3_6_pmc_get_ctrl_settings(adev,
config,
&lo_base_addr,
&hi_base_addr,
NULL,
NULL);
if (xgmi_tx_link >= 0) {
ret = df_v3_6_stop_xgmi_link_cntr(adev,
xgmi_tx_link,
is_disable);
if (ret)
return ret;
}
if (ret)
return ret;
ret = 0;
break;
lo_val = RREG32_PCIE(lo_base_addr);
DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x",
config, lo_base_addr, hi_base_addr, lo_val);
WREG32_PCIE(lo_base_addr, lo_val & ~(1ULL << 22));
if (is_disable)
df_v3_6_pmc_release_cntr(adev, config);
break;
default:
break;
}
@ -510,28 +459,38 @@ static void df_v3_6_pmc_get_count(struct amdgpu_device *adev,
uint64_t config,
uint64_t *count)
{
int xgmi_tx_link;
uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
*count = 0;
switch (adev->asic_type) {
case CHIP_VEGA20:
xgmi_tx_link = IS_DF_XGMI_0_TX(config) ? 0
: (IS_DF_XGMI_1_TX(config) ? 1 : -1);
if (xgmi_tx_link >= 0) {
df_v3_6_reset_xgmi_link_cntr(adev, xgmi_tx_link);
df_v3_6_get_xgmi_link_cntr(adev, xgmi_tx_link, count);
}
df_v3_6_pmc_get_read_settings(adev, config, &lo_base_addr,
&hi_base_addr);
if ((lo_base_addr == 0) || (hi_base_addr == 0))
return;
lo_val = RREG32_PCIE(lo_base_addr);
hi_val = RREG32_PCIE(hi_base_addr);
*count = ((hi_val | 0ULL) << 32) | (lo_val | 0ULL);
if (*count >= DF_V3_6_PERFMON_OVERFLOW)
*count = 0;
DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x:%08x",
config, lo_base_addr, hi_base_addr, lo_val, hi_val);
break;
default:
break;
}
}
const struct amdgpu_df_funcs df_v3_6_funcs = {
.init = df_v3_6_init,
.sw_init = df_v3_6_sw_init,
.enable_broadcast_mode = df_v3_6_enable_broadcast_mode,
.get_fb_channel_number = df_v3_6_get_fb_channel_number,
.get_hbm_channel_number = df_v3_6_get_hbm_channel_number,

19
drivers/gpu/drm/amd/amdgpu/df_v3_6.h
Просмотреть файл

@ -36,22 +36,15 @@ enum DF_V3_6_MGCG {
};
/* Defined in global_features.h as FTI_PERFMON_VISIBLE */
#define AMDGPU_DF_MAX_COUNTERS 4
#define DF_V3_6_MAX_COUNTERS 4
/* get flags from df perfmon config */
#define GET_EVENT(x) (x & 0xFFUL)
#define GET_INSTANCE(x) ((x >> 8) & 0xFFUL)
#define GET_UNITMASK(x) ((x >> 16) & 0xFFUL)
#define GET_INSTANCE_CONFIG(x) (0ULL | (0x07ULL) \
| ((0x046ULL + x) << 8) \
| (0x02 << 16))
/* df event conf macros */
#define IS_DF_XGMI_0_TX(x) (GET_EVENT(x) == 0x7 \
&& GET_INSTANCE(x) == 0x46 && GET_UNITMASK(x) == 0x2)
#define IS_DF_XGMI_1_TX(x) (GET_EVENT(x) == 0x7 \
&& GET_INSTANCE(x) == 0x47 && GET_UNITMASK(x) == 0x2)
#define DF_V3_6_GET_EVENT(x) (x & 0xFFUL)
#define DF_V3_6_GET_INSTANCE(x) ((x >> 8) & 0xFFUL)
#define DF_V3_6_GET_UNITMASK(x) ((x >> 16) & 0xFFUL)
#define DF_V3_6_PERFMON_OVERFLOW 0xFFFFFFFFFFFFULL
extern const struct attribute_group *df_v3_6_attr_groups[];
extern const struct amdgpu_df_funcs df_v3_6_funcs;
#endif

5216
drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c Normal file
Просмотреть файл

Разница между файлами не показана из-за своего большого размера Загрузить разницу

29
drivers/gpu/drm/amd/amdgpu/gfx_v10_0.h Normal file
Просмотреть файл

@ -0,0 +1,29 @@
/*
* Copyright 2019 dvanced 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 __GFX_V10_0_H__
#define __GFX_V10_0_H__
extern const struct amdgpu_ip_block_version gfx_v10_0_ip_block;
#endif

4
drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
Просмотреть файл

@ -3115,7 +3115,7 @@ static int gfx_v6_0_sw_init(void *handle)
ring->ring_obj = NULL;
sprintf(ring->name, "gfx");
r = amdgpu_ring_init(adev, ring, 1024,
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP);
if (r)
return r;
}
@ -3350,7 +3350,7 @@ static int gfx_v6_0_set_eop_interrupt_state(struct amdgpu_device *adev,
enum amdgpu_interrupt_state state)
{
switch (type) {
case AMDGPU_CP_IRQ_GFX_EOP:
case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
gfx_v6_0_set_gfx_eop_interrupt_state(adev, state);
break;
case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:

4
drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
Просмотреть файл

@ -4462,7 +4462,7 @@ static int gfx_v7_0_sw_init(void *handle)
ring->ring_obj = NULL;
sprintf(ring->name, "gfx");
r = amdgpu_ring_init(adev, ring, 1024,
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP);
if (r)
return r;
}
@ -4799,7 +4799,7 @@ static int gfx_v7_0_set_eop_interrupt_state(struct amdgpu_device *adev,
enum amdgpu_interrupt_state state)
{
switch (type) {
case AMDGPU_CP_IRQ_GFX_EOP:
case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
gfx_v7_0_set_gfx_eop_interrupt_state(adev, state);
break;
case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:

26
drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
Просмотреть файл

@ -2009,7 +2009,7 @@ static int gfx_v8_0_sw_init(void *handle)
}
r = amdgpu_ring_init(adev, ring, 1024, &adev->gfx.eop_irq,
AMDGPU_CP_IRQ_GFX_EOP);
AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP);
if (r)
return r;
}
@ -2046,7 +2046,7 @@ static int gfx_v8_0_sw_init(void *handle)
return r;
/* create MQD for all compute queues as well as KIQ for SRIOV case */
r = amdgpu_gfx_compute_mqd_sw_init(adev, sizeof(struct vi_mqd_allocation));
r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct vi_mqd_allocation));
if (r)
return r;
@ -2069,7 +2069,7 @@ static int gfx_v8_0_sw_fini(void *handle)
for (i = 0; i < adev->gfx.num_compute_rings; i++)
amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
amdgpu_gfx_compute_mqd_sw_fini(adev);
amdgpu_gfx_mqd_sw_fini(adev);
amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq.ring, &adev->gfx.kiq.irq);
amdgpu_gfx_kiq_fini(adev);
@ -6217,7 +6217,7 @@ static void gfx_v8_0_pipe_reserve_resources(struct amdgpu_device *adev,
struct amdgpu_ring *iring;
mutex_lock(&adev->gfx.pipe_reserve_mutex);
pipe = amdgpu_gfx_queue_to_bit(adev, ring->me, ring->pipe, 0);
pipe = amdgpu_gfx_mec_queue_to_bit(adev, ring->me, ring->pipe, 0);
if (acquire)
set_bit(pipe, adev->gfx.pipe_reserve_bitmap);
else
@ -6236,20 +6236,20 @@ static void gfx_v8_0_pipe_reserve_resources(struct amdgpu_device *adev,
/* Lower all pipes without a current reservation */
for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
iring = &adev->gfx.gfx_ring[i];
pipe = amdgpu_gfx_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
pipe = amdgpu_gfx_mec_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
gfx_v8_0_ring_set_pipe_percent(iring, reserve);
}
for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
iring = &adev->gfx.compute_ring[i];
pipe = amdgpu_gfx_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
pipe = amdgpu_gfx_mec_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
gfx_v8_0_ring_set_pipe_percent(iring, reserve);
}
@ -6537,7 +6537,7 @@ static int gfx_v8_0_set_eop_interrupt_state(struct amdgpu_device *adev,
enum amdgpu_interrupt_state state)
{
switch (type) {
case AMDGPU_CP_IRQ_GFX_EOP:
case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
gfx_v8_0_set_gfx_eop_interrupt_state(adev, state);
break;
case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:

164
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
Просмотреть файл

@ -309,6 +309,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
static uint64_t gfx_v9_0_get_gpu_clock_counter(struct amdgpu_device *adev);
static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance);
static void gfx_v9_0_ring_emit_de_meta(struct amdgpu_ring *ring);
static u64 gfx_v9_0_ring_get_rptr_compute(struct amdgpu_ring *ring);
static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev)
{
@ -1724,7 +1725,7 @@ static int gfx_v9_0_sw_init(void *handle)
ring->use_doorbell = true;
ring->doorbell_index = adev->doorbell_index.gfx_ring0 << 1;
r = amdgpu_ring_init(adev, ring, 1024,
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_EOP);
&adev->gfx.eop_irq, AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP);
if (r)
return r;
}
@ -1760,7 +1761,7 @@ static int gfx_v9_0_sw_init(void *handle)
return r;
/* create MQD for all compute queues as wel as KIQ for SRIOV case */
r = amdgpu_gfx_compute_mqd_sw_init(adev, sizeof(struct v9_mqd_allocation));
r = amdgpu_gfx_mqd_sw_init(adev, sizeof(struct v9_mqd_allocation));
if (r)
return r;
@ -1802,7 +1803,7 @@ static int gfx_v9_0_sw_fini(void *handle)
for (i = 0; i < adev->gfx.num_compute_rings; i++)
amdgpu_ring_fini(&adev->gfx.compute_ring[i]);
amdgpu_gfx_compute_mqd_sw_fini(adev);
amdgpu_gfx_mqd_sw_fini(adev);
amdgpu_gfx_kiq_free_ring(&adev->gfx.kiq.ring, &adev->gfx.kiq.irq);
amdgpu_gfx_kiq_fini(adev);
@ -3600,45 +3601,89 @@ static const struct soc15_reg_entry sgpr_init_regs[] = {
};
static const struct soc15_reg_entry sec_ded_counter_registers[] = {
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_DMA_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_TAG_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_DED) },
{ SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_DED_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_INFO) },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_SEC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmTCP_ATC_EDC_GATCL1_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2) },
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2) },
{ SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2) },
{ SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT) },
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_SCRATCH_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPC_EDC_UCODE_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_ROQ_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPF_EDC_TAG_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_DMA_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmCPG_EDC_TAG_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_CSINVOC_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_RESTORE_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmDC_EDC_STATE_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_GRBM_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_DED), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmSPI_EDC_CNT), 0, 4, 1},
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT), 0, 4, 6},
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_DED_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_INFO), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_SEC_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT), 0, 1, 16},
{ SOC15_REG_ENTRY(GC, 0, mmTCP_ATC_EDC_GATCL1_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmTCP_EDC_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmTD_EDC_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT2), 0, 4, 6},
{ SOC15_REG_ENTRY(GC, 0, mmSQ_EDC_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmTA_EDC_CNT), 0, 4, 16},
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PHY_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmGDS_EDC_OA_PIPE_CNT), 0, 1, 1},
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT), 0, 1, 32},
{ SOC15_REG_ENTRY(GC, 0, mmGCEA_EDC_CNT2), 0, 1, 32},
{ SOC15_REG_ENTRY(GC, 0, mmTCI_EDC_CNT), 0, 1, 72},
{ SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2), 0, 1, 16},
{ SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT), 0, 1, 2},
{ SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6},
};
static int gfx_v9_0_do_edc_gds_workarounds(struct amdgpu_device *adev)
{
struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
int i, r;
r = amdgpu_ring_alloc(ring, 7);
if (r) {
DRM_ERROR("amdgpu: GDS workarounds failed to lock ring %s (%d).\n",
ring->name, r);
return r;
}
WREG32_SOC15(GC, 0, mmGDS_VMID0_BASE, 0x00000000);
WREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE, adev->gds.gds_size);
amdgpu_ring_write(ring, PACKET3(PACKET3_DMA_DATA, 5));
amdgpu_ring_write(ring, (PACKET3_DMA_DATA_CP_SYNC |
PACKET3_DMA_DATA_DST_SEL(1) |
PACKET3_DMA_DATA_SRC_SEL(2) |
PACKET3_DMA_DATA_ENGINE(0)));
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, 0);
amdgpu_ring_write(ring, PACKET3_DMA_DATA_CMD_RAW_WAIT |
adev->gds.gds_size);
amdgpu_ring_commit(ring);
for (i = 0; i < adev->usec_timeout; i++) {
if (ring->wptr == gfx_v9_0_ring_get_rptr_compute(ring))
break;
udelay(1);
}
if (i >= adev->usec_timeout)
r = -ETIMEDOUT;
WREG32_SOC15(GC, 0, mmGDS_VMID0_SIZE, 0x00000000);
return r;
}
static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
{
struct amdgpu_ring *ring = &adev->gfx.compute_ring[0];
struct amdgpu_ib ib;
struct dma_fence *f = NULL;
int r, i, j;
int r, i, j, k;
unsigned total_size, vgpr_offset, sgpr_offset;
u64 gpu_addr;
@ -3750,19 +3795,13 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
/* read back registers to clear the counters */
mutex_lock(&adev->grbm_idx_mutex);
for (j = 0; j < 16; j++) {
gfx_v9_0_select_se_sh(adev, 0x01, 0x0, j);
for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++)
RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
gfx_v9_0_select_se_sh(adev, 0x02, 0x0, j);
for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++)
RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
gfx_v9_0_select_se_sh(adev, 0x03, 0x0, j);
for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++)
RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
gfx_v9_0_select_se_sh(adev, 0x04, 0x0, j);
for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++)
RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++) {
for (j = 0; j < sec_ded_counter_registers[i].se_num; j++) {
for (k = 0; k < sec_ded_counter_registers[i].instance; k++) {
gfx_v9_0_select_se_sh(adev, j, 0x0, k);
RREG32(SOC15_REG_ENTRY_OFFSET(sec_ded_counter_registers[i]));
}
}
}
WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, 0xe0000000);
mutex_unlock(&adev->grbm_idx_mutex);
@ -3815,6 +3854,10 @@ static int gfx_v9_0_ecc_late_init(void *handle)
return 0;
}
r = gfx_v9_0_do_edc_gds_workarounds(adev);
if (r)
return r;
/* requires IBs so do in late init after IB pool is initialized */
r = gfx_v9_0_do_edc_gpr_workarounds(adev);
if (r)
@ -3864,9 +3907,7 @@ static int gfx_v9_0_ecc_late_init(void *handle)
if (r)
goto interrupt;
r = amdgpu_ras_debugfs_create(adev, &fs_info);
if (r)
goto debugfs;
amdgpu_ras_debugfs_create(adev, &fs_info);
r = amdgpu_ras_sysfs_create(adev, &fs_info);
if (r)
@ -3881,7 +3922,6 @@ irq:
amdgpu_ras_sysfs_remove(adev, *ras_if);
sysfs:
amdgpu_ras_debugfs_remove(adev, *ras_if);
debugfs:
amdgpu_ras_interrupt_remove_handler(adev, &ih_info);
interrupt:
amdgpu_ras_feature_enable(adev, *ras_if, 0);
@ -4542,7 +4582,7 @@ static void gfx_v9_0_pipe_reserve_resources(struct amdgpu_device *adev,
struct amdgpu_ring *iring;
mutex_lock(&adev->gfx.pipe_reserve_mutex);
pipe = amdgpu_gfx_queue_to_bit(adev, ring->me, ring->pipe, 0);
pipe = amdgpu_gfx_mec_queue_to_bit(adev, ring->me, ring->pipe, 0);
if (acquire)
set_bit(pipe, adev->gfx.pipe_reserve_bitmap);
else
@ -4561,20 +4601,20 @@ static void gfx_v9_0_pipe_reserve_resources(struct amdgpu_device *adev,
/* Lower all pipes without a current reservation */
for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
iring = &adev->gfx.gfx_ring[i];
pipe = amdgpu_gfx_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
pipe = amdgpu_gfx_mec_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
gfx_v9_0_ring_set_pipe_percent(iring, reserve);
}
for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
iring = &adev->gfx.compute_ring[i];
pipe = amdgpu_gfx_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
pipe = amdgpu_gfx_mec_queue_to_bit(adev,
iring->me,
iring->pipe,
0);
reserve = test_bit(pipe, adev->gfx.pipe_reserve_bitmap);
gfx_v9_0_ring_set_pipe_percent(iring, reserve);
}
@ -4989,7 +5029,7 @@ static int gfx_v9_0_set_eop_interrupt_state(struct amdgpu_device *adev,
enum amdgpu_interrupt_state state)
{
switch (type) {
case AMDGPU_CP_IRQ_GFX_EOP:
case AMDGPU_CP_IRQ_GFX_ME0_PIPE0_EOP:
gfx_v9_0_set_gfx_eop_interrupt_state(adev, state);
break;
case AMDGPU_CP_IRQ_COMPUTE_MEC1_PIPE0_EOP:

353
drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.c Normal file
Просмотреть файл

@ -0,0 +1,353 @@
/*
* Copyright 2019 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.
*
*/
#include "amdgpu.h"
#include "gfxhub_v2_0.h"
#include "gc/gc_10_1_0_offset.h"
#include "gc/gc_10_1_0_sh_mask.h"
#include "gc/gc_10_1_0_default.h"
#include "navi10_enum.h"
#include "soc15_common.h"
u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev)
{
u64 base = RREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE);
base &= GCMC_VM_FB_LOCATION_BASE__FB_BASE_MASK;
base <<= 24;
return base;
}
u64 gfxhub_v2_0_get_mc_fb_offset(struct amdgpu_device *adev)
{
return (u64)RREG32_SOC15(GC, 0, mmGCMC_VM_FB_OFFSET) << 24;
}
static void gfxhub_v2_0_init_gart_pt_regs(struct amdgpu_device *adev)
{
uint64_t value = amdgpu_gmc_pd_addr(adev->gart.bo);
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
lower_32_bits(value));
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
upper_32_bits(value));
}
static void gfxhub_v2_0_init_gart_aperture_regs(struct amdgpu_device *adev)
{
gfxhub_v2_0_init_gart_pt_regs(adev);
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
(u32)(adev->gmc.gart_start >> 12));
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
(u32)(adev->gmc.gart_start >> 44));
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
(u32)(adev->gmc.gart_end >> 12));
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
(u32)(adev->gmc.gart_end >> 44));
}
static void gfxhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev)
{
uint64_t value;
/* Disable AGP. */
WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BASE, 0);
WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_TOP, 0);
WREG32_SOC15(GC, 0, mmGCMC_VM_AGP_BOT, 0x00FFFFFF);
/* Program the system aperture low logical page number. */
WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_LOW_ADDR,
adev->gmc.vram_start >> 18);
WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
adev->gmc.vram_end >> 18);
/* Set default page address. */
value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start
+ adev->vm_manager.vram_base_offset;
WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
(u32)(value >> 12));
WREG32_SOC15(GC, 0, mmGCMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
(u32)(value >> 44));
/* Program "protection fault". */
WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
(u32)(adev->dummy_page_addr >> 12));
WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
(u32)((u64)adev->dummy_page_addr >> 44));
WREG32_FIELD15(GC, 0, GCVM_L2_PROTECTION_FAULT_CNTL2,
ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
}
static void gfxhub_v2_0_init_tlb_regs(struct amdgpu_device *adev)
{
uint32_t tmp;
/* Setup TLB control */
tmp = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
ENABLE_ADVANCED_DRIVER_MODEL, 1);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
MTYPE, MTYPE_UC); /* UC, uncached */
WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp);
}
static void gfxhub_v2_0_init_cache_regs(struct amdgpu_device *adev)
{
uint32_t tmp;
/* Setup L2 cache */
tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_CACHE, 1);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
/* XXX for emulation, Refer to closed source code.*/
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL,
L2_PDE0_CACHE_TAG_GENERATION_MODE, 0);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 1);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL, tmp);
tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL2, tmp);
tmp = mmGCVM_L2_CNTL3_DEFAULT;
WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, tmp);
tmp = mmGCVM_L2_CNTL4_DEFAULT;
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
tmp = REG_SET_FIELD(tmp, GCVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL4, tmp);
}
static void gfxhub_v2_0_enable_system_domain(struct amdgpu_device *adev)
{
uint32_t tmp;
tmp = RREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_CNTL);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
WREG32_SOC15(GC, 0, mmGCVM_CONTEXT0_CNTL, tmp);
}
static void gfxhub_v2_0_disable_identity_aperture(struct amdgpu_device *adev)
{
WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
0xFFFFFFFF);
WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
0x0000000F);
WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32,
0);
WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32,
0);
WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32, 0);
WREG32_SOC15(GC, 0, mmGCVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32, 0);
}
static void gfxhub_v2_0_setup_vmid_config(struct amdgpu_device *adev)
{
int i;
uint32_t tmp;
for (i = 0; i <= 14; i++) {
tmp = RREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL, i);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
adev->vm_manager.num_level);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
PAGE_TABLE_BLOCK_SIZE,
adev->vm_manager.block_size - 9);
/* Send no-retry XNACK on fault to suppress VM fault storm. */
tmp = REG_SET_FIELD(tmp, GCVM_CONTEXT1_CNTL,
RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_CNTL, i, tmp);
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32, i*2,
lower_32_bits(adev->vm_manager.max_pfn - 1));
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, i*2,
upper_32_bits(adev->vm_manager.max_pfn - 1));
}
}
static void gfxhub_v2_0_program_invalidation(struct amdgpu_device *adev)
{
unsigned i;
for (i = 0 ; i < 18; ++i) {
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
2 * i, 0xffffffff);
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
2 * i, 0x1f);
}
}
int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev)
{
if (amdgpu_sriov_vf(adev)) {
/*
* GCMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
* VF copy registers so vbios post doesn't program them, for
* SRIOV driver need to program them
*/
WREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_BASE,
adev->gmc.vram_start >> 24);
WREG32_SOC15(GC, 0, mmGCMC_VM_FB_LOCATION_TOP,
adev->gmc.vram_end >> 24);
}
/* GART Enable. */
gfxhub_v2_0_init_gart_aperture_regs(adev);
gfxhub_v2_0_init_system_aperture_regs(adev);
gfxhub_v2_0_init_tlb_regs(adev);
gfxhub_v2_0_init_cache_regs(adev);
gfxhub_v2_0_enable_system_domain(adev);
gfxhub_v2_0_disable_identity_aperture(adev);
gfxhub_v2_0_setup_vmid_config(adev);
gfxhub_v2_0_program_invalidation(adev);
return 0;
}
void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev)
{
u32 tmp;
u32 i;
/* Disable all tables */
for (i = 0; i < 16; i++)
WREG32_SOC15_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL, i, 0);
/* Setup TLB control */
tmp = RREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
tmp = REG_SET_FIELD(tmp, GCMC_VM_MX_L1_TLB_CNTL,
ENABLE_ADVANCED_DRIVER_MODEL, 0);
WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp);
/* Setup L2 cache */
WREG32_FIELD15(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0);
WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, 0);
}
/**
* gfxhub_v2_0_set_fault_enable_default - update GART/VM fault handling
*
* @adev: amdgpu_device pointer
* @value: true redirects VM faults to the default page
*/
void gfxhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev,
bool value)
{
u32 tmp;
tmp = RREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
if (!value) {
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
CRASH_ON_NO_RETRY_FAULT, 1);
tmp = REG_SET_FIELD(tmp, GCVM_L2_PROTECTION_FAULT_CNTL,
CRASH_ON_RETRY_FAULT, 1);
}
WREG32_SOC15(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL, tmp);
}
void gfxhub_v2_0_init(struct amdgpu_device *adev)
{
struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_GFXHUB];
hub->ctx0_ptb_addr_lo32 =
SOC15_REG_OFFSET(GC, 0,
mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
hub->ctx0_ptb_addr_hi32 =
SOC15_REG_OFFSET(GC, 0,
mmGCVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
hub->vm_inv_eng0_req =
SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_REQ);
hub->vm_inv_eng0_ack =
SOC15_REG_OFFSET(GC, 0, mmGCVM_INVALIDATE_ENG0_ACK);
hub->vm_context0_cntl =
SOC15_REG_OFFSET(GC, 0, mmGCVM_CONTEXT0_CNTL);
hub->vm_l2_pro_fault_status =
SOC15_REG_OFFSET(GC, 0, mmGCVM_L2_PROTECTION_FAULT_STATUS);
hub->vm_l2_pro_fault_cntl =
SOC15_REG_OFFSET(GC, 0, mmGCVM_L2_PROTECTION_FAULT_CNTL);
}

35
drivers/gpu/drm/amd/amdgpu/gfxhub_v2_0.h Normal file
Просмотреть файл

@ -0,0 +1,35 @@
/*
* Copyright 2019 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 __GFXHUB_V2_0_H__
#define __GFXHUB_V2_0_H__
u64 gfxhub_v2_0_get_fb_location(struct amdgpu_device *adev);
int gfxhub_v2_0_gart_enable(struct amdgpu_device *adev);
void gfxhub_v2_0_gart_disable(struct amdgpu_device *adev);
void gfxhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev,
bool value);
void gfxhub_v2_0_init(struct amdgpu_device *adev);
u64 gfxhub_v2_0_get_mc_fb_offset(struct amdgpu_device *adev);
#endif

917
drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c Normal file
Просмотреть файл

@ -0,0 +1,917 @@
/*
* Copyright 2019 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.
*
*/
#include <linux/firmware.h>
#include <linux/pci.h>
#include "amdgpu.h"
#include "amdgpu_atomfirmware.h"
#include "gmc_v10_0.h"
#include "hdp/hdp_5_0_0_offset.h"
#include "hdp/hdp_5_0_0_sh_mask.h"
#include "gc/gc_10_1_0_sh_mask.h"
#include "mmhub/mmhub_2_0_0_sh_mask.h"
#include "dcn/dcn_2_0_0_offset.h"
#include "dcn/dcn_2_0_0_sh_mask.h"
#include "oss/osssys_5_0_0_offset.h"
#include "ivsrcid/vmc/irqsrcs_vmc_1_0.h"
#include "navi10_enum.h"
#include "soc15.h"
#include "soc15_common.h"
#include "nbio_v2_3.h"
#include "gfxhub_v2_0.h"
#include "mmhub_v2_0.h"
#include "athub_v2_0.h"
/* XXX Move this macro to navi10 header file, which is like vid.h for VI.*/
#define AMDGPU_NUM_OF_VMIDS 8
#if 0
static const struct soc15_reg_golden golden_settings_navi10_hdp[] =
{
/* TODO add golden setting for hdp */
};
#endif
static int
gmc_v10_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
struct amdgpu_irq_src *src, unsigned type,
enum amdgpu_interrupt_state state)
{
struct amdgpu_vmhub *hub;
u32 tmp, reg, bits[AMDGPU_MAX_VMHUBS], i;
bits[AMDGPU_GFXHUB] = GCVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
GCVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
GCVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
GCVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
GCVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
GCVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
GCVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
bits[AMDGPU_MMHUB] = MMVM_CONTEXT1_CNTL__RANGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
MMVM_CONTEXT1_CNTL__DUMMY_PAGE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
MMVM_CONTEXT1_CNTL__PDE0_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
MMVM_CONTEXT1_CNTL__VALID_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
MMVM_CONTEXT1_CNTL__READ_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
MMVM_CONTEXT1_CNTL__WRITE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK |
MMVM_CONTEXT1_CNTL__EXECUTE_PROTECTION_FAULT_ENABLE_INTERRUPT_MASK;
switch (state) {
case AMDGPU_IRQ_STATE_DISABLE:
/* MM HUB */
hub = &adev->vmhub[AMDGPU_MMHUB];
for (i = 0; i < 16; i++) {
reg = hub->vm_context0_cntl + i;
tmp = RREG32(reg);
tmp &= ~bits[AMDGPU_MMHUB];
WREG32(reg, tmp);
}
/* GFX HUB */
hub = &adev->vmhub[AMDGPU_GFXHUB];
for (i = 0; i < 16; i++) {
reg = hub->vm_context0_cntl + i;
tmp = RREG32(reg);
tmp &= ~bits[AMDGPU_GFXHUB];
WREG32(reg, tmp);
}
break;
case AMDGPU_IRQ_STATE_ENABLE:
/* MM HUB */
hub = &adev->vmhub[AMDGPU_MMHUB];
for (i = 0; i < 16; i++) {
reg = hub->vm_context0_cntl + i;
tmp = RREG32(reg);
tmp |= bits[AMDGPU_MMHUB];
WREG32(reg, tmp);
}
/* GFX HUB */
hub = &adev->vmhub[AMDGPU_GFXHUB];
for (i = 0; i < 16; i++) {
reg = hub->vm_context0_cntl + i;
tmp = RREG32(reg);
tmp |= bits[AMDGPU_GFXHUB];
WREG32(reg, tmp);
}
break;
default:
break;
}
return 0;
}
static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
{
struct amdgpu_vmhub *hub = &adev->vmhub[entry->vmid_src];
uint32_t status = 0;
u64 addr;
addr = (u64)entry->src_data[0] << 12;
addr |= ((u64)entry->src_data[1] & 0xf) << 44;
if (!amdgpu_sriov_vf(adev)) {
status = RREG32(hub->vm_l2_pro_fault_status);
WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
}
if (printk_ratelimit()) {
dev_err(adev->dev,
"[%s] VMC page fault (src_id:%u ring:%u vmid:%u pasid:%u)\n",
entry->vmid_src ? "mmhub" : "gfxhub",
entry->src_id, entry->ring_id, entry->vmid,
entry->pasid);
dev_err(adev->dev, " at page 0x%016llx from %d\n",
addr, entry->client_id);
if (!amdgpu_sriov_vf(adev))
dev_err(adev->dev,
"VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n",
status);
}
return 0;
}
static const struct amdgpu_irq_src_funcs gmc_v10_0_irq_funcs = {
.set = gmc_v10_0_vm_fault_interrupt_state,
.process = gmc_v10_0_process_interrupt,
};
static void gmc_v10_0_set_irq_funcs(struct amdgpu_device *adev)
{
adev->gmc.vm_fault.num_types = 1;
adev->gmc.vm_fault.funcs = &gmc_v10_0_irq_funcs;
}
static uint32_t gmc_v10_0_get_invalidate_req(unsigned int vmid,
uint32_t flush_type)
{
u32 req = 0;
/* invalidate using legacy mode on vmid*/
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
PER_VMID_INVALIDATE_REQ, 1 << vmid);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, FLUSH_TYPE, flush_type);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PTES, 1);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE0, 1);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE1, 1);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L2_PDE2, 1);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ, INVALIDATE_L1_PTES, 1);
req = REG_SET_FIELD(req, GCVM_INVALIDATE_ENG0_REQ,
CLEAR_PROTECTION_FAULT_STATUS_ADDR, 0);
return req;
}
/*
* GART
* VMID 0 is the physical GPU addresses as used by the kernel.
* VMIDs 1-15 are used for userspace clients and are handled
* by the amdgpu vm/hsa code.
*/
static void gmc_v10_0_flush_vm_hub(struct amdgpu_device *adev, uint32_t vmid,
unsigned int vmhub, uint32_t flush_type)
{
struct amdgpu_vmhub *hub = &adev->vmhub[vmhub];
u32 tmp = gmc_v10_0_get_invalidate_req(vmid, flush_type);
/* Use register 17 for GART */
const unsigned eng = 17;
unsigned int i;
WREG32_NO_KIQ(hub->vm_inv_eng0_req + eng, tmp);
/* Wait for ACK with a delay.*/
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32_NO_KIQ(hub->vm_inv_eng0_ack + eng);
tmp &= 1 << vmid;
if (tmp)
break;
udelay(1);
}
if (i < adev->usec_timeout)
return;
DRM_ERROR("Timeout waiting for VM flush ACK!\n");
}
/**
* gmc_v10_0_flush_gpu_tlb - gart tlb flush callback
*
* @adev: amdgpu_device pointer
* @vmid: vm instance to flush
*
* Flush the TLB for the requested page table.
*/
static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev,
uint32_t vmid, uint32_t flush_type)
{
struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
struct dma_fence *fence;
struct amdgpu_job *job;
int r;
/* flush hdp cache */
adev->nbio_funcs->hdp_flush(adev, NULL);
mutex_lock(&adev->mman.gtt_window_lock);
gmc_v10_0_flush_vm_hub(adev, vmid, AMDGPU_MMHUB, 0);
if (!adev->mman.buffer_funcs_enabled || !adev->ib_pool_ready ||
adev->asic_type != CHIP_NAVI10) {
gmc_v10_0_flush_vm_hub(adev, vmid, AMDGPU_GFXHUB, 0);
mutex_unlock(&adev->mman.gtt_window_lock);
return;
}
/* The SDMA on Navi has a bug which can theoretically result in memory
* corruption if an invalidation happens at the same time as an VA
* translation. Avoid this by doing the invalidation from the SDMA
* itself.
*/
r = amdgpu_job_alloc_with_ib(adev, 16 * 4, &job);
if (r)
goto error_alloc;
job->vm_pd_addr = amdgpu_gmc_pd_addr(adev->gart.bo);
job->vm_needs_flush = true;
amdgpu_ring_pad_ib(ring, &job->ibs[0]);
r = amdgpu_job_submit(job, &adev->mman.entity,
AMDGPU_FENCE_OWNER_UNDEFINED, &fence);
if (r)
goto error_submit;
mutex_unlock(&adev->mman.gtt_window_lock);
dma_fence_wait(fence, false);
dma_fence_put(fence);
return;
error_submit:
amdgpu_job_free(job);
error_alloc:
mutex_unlock(&adev->mman.gtt_window_lock);
DRM_ERROR("Error flushing GPU TLB using the SDMA (%d)!\n", r);
}
static uint64_t gmc_v10_0_emit_flush_gpu_tlb(struct amdgpu_ring *ring,
unsigned vmid, uint64_t pd_addr)
{
struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub];
uint32_t req = gmc_v10_0_get_invalidate_req(vmid, 0);
unsigned eng = ring->vm_inv_eng;
amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_lo32 + (2 * vmid),
lower_32_bits(pd_addr));
amdgpu_ring_emit_wreg(ring, hub->ctx0_ptb_addr_hi32 + (2 * vmid),
upper_32_bits(pd_addr));
amdgpu_ring_emit_wreg(ring, hub->vm_inv_eng0_req + eng, req);
/* wait for the invalidate to complete */
amdgpu_ring_emit_reg_wait(ring, hub->vm_inv_eng0_ack + eng,
1 << vmid, 1 << vmid);
return pd_addr;
}
static void gmc_v10_0_emit_pasid_mapping(struct amdgpu_ring *ring, unsigned vmid,
unsigned pasid)
{
struct amdgpu_device *adev = ring->adev;
uint32_t reg;
if (ring->funcs->vmhub == AMDGPU_GFXHUB)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT) + vmid;
else
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_VMID_0_LUT_MM) + vmid;
amdgpu_ring_emit_wreg(ring, reg, pasid);
}
/*
* PTE format on NAVI 10:
* 63:59 reserved
* 58:57 reserved
* 56 F
* 55 L
* 54 reserved
* 53:52 SW
* 51 T
* 50:48 mtype
* 47:12 4k physical page base address
* 11:7 fragment
* 6 write
* 5 read
* 4 exe
* 3 Z
* 2 snooped
* 1 system
* 0 valid
*
* PDE format on NAVI 10:
* 63:59 block fragment size
* 58:55 reserved
* 54 P
* 53:48 reserved
* 47:6 physical base address of PD or PTE
* 5:3 reserved
* 2 C
* 1 system
* 0 valid
*/
static uint64_t gmc_v10_0_get_vm_pte_flags(struct amdgpu_device *adev,
uint32_t flags)
{
uint64_t pte_flag = 0;
if (flags & AMDGPU_VM_PAGE_EXECUTABLE)
pte_flag |= AMDGPU_PTE_EXECUTABLE;
if (flags & AMDGPU_VM_PAGE_READABLE)
pte_flag |= AMDGPU_PTE_READABLE;
if (flags & AMDGPU_VM_PAGE_WRITEABLE)
pte_flag |= AMDGPU_PTE_WRITEABLE;
switch (flags & AMDGPU_VM_MTYPE_MASK) {
case AMDGPU_VM_MTYPE_DEFAULT:
pte_flag |= AMDGPU_PTE_MTYPE_NV10(MTYPE_NC);
break;
case AMDGPU_VM_MTYPE_NC:
pte_flag |= AMDGPU_PTE_MTYPE_NV10(MTYPE_NC);
break;
case AMDGPU_VM_MTYPE_WC:
pte_flag |= AMDGPU_PTE_MTYPE_NV10(MTYPE_WC);
break;
case AMDGPU_VM_MTYPE_CC:
pte_flag |= AMDGPU_PTE_MTYPE_NV10(MTYPE_CC);
break;
case AMDGPU_VM_MTYPE_UC:
pte_flag |= AMDGPU_PTE_MTYPE_NV10(MTYPE_UC);
break;
default:
pte_flag |= AMDGPU_PTE_MTYPE_NV10(MTYPE_NC);
break;
}
if (flags & AMDGPU_VM_PAGE_PRT)
pte_flag |= AMDGPU_PTE_PRT;
return pte_flag;
}
static void gmc_v10_0_get_vm_pde(struct amdgpu_device *adev, int level,
uint64_t *addr, uint64_t *flags)
{
if (!(*flags & AMDGPU_PDE_PTE) && !(*flags & AMDGPU_PTE_SYSTEM))
*addr = adev->vm_manager.vram_base_offset + *addr -
adev->gmc.vram_start;
BUG_ON(*addr & 0xFFFF00000000003FULL);
if (!adev->gmc.translate_further)
return;
if (level == AMDGPU_VM_PDB1) {
/* Set the block fragment size */
if (!(*flags & AMDGPU_PDE_PTE))
*flags |= AMDGPU_PDE_BFS(0x9);
} else if (level == AMDGPU_VM_PDB0) {
if (*flags & AMDGPU_PDE_PTE)
*flags &= ~AMDGPU_PDE_PTE;
else
*flags |= AMDGPU_PTE_TF;
}
}
static const struct amdgpu_gmc_funcs gmc_v10_0_gmc_funcs = {
.flush_gpu_tlb = gmc_v10_0_flush_gpu_tlb,
.emit_flush_gpu_tlb = gmc_v10_0_emit_flush_gpu_tlb,
.emit_pasid_mapping = gmc_v10_0_emit_pasid_mapping,
.get_vm_pte_flags = gmc_v10_0_get_vm_pte_flags,
.get_vm_pde = gmc_v10_0_get_vm_pde
};
static void gmc_v10_0_set_gmc_funcs(struct amdgpu_device *adev)
{
if (adev->gmc.gmc_funcs == NULL)
adev->gmc.gmc_funcs = &gmc_v10_0_gmc_funcs;
}
static int gmc_v10_0_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
gmc_v10_0_set_gmc_funcs(adev);
gmc_v10_0_set_irq_funcs(adev);
adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
adev->gmc.shared_aperture_end =
adev->gmc.shared_aperture_start + (4ULL << 30) - 1;
adev->gmc.private_aperture_start = 0x1000000000000000ULL;
adev->gmc.private_aperture_end =
adev->gmc.private_aperture_start + (4ULL << 30) - 1;
return 0;
}
static int gmc_v10_0_late_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
unsigned vm_inv_eng[AMDGPU_MAX_VMHUBS] = { 4, 4 };
unsigned i;
for(i = 0; i < adev->num_rings; ++i) {
struct amdgpu_ring *ring = adev->rings[i];
unsigned vmhub = ring->funcs->vmhub;
ring->vm_inv_eng = vm_inv_eng[vmhub]++;
dev_info(adev->dev, "ring %u(%s) uses VM inv eng %u on hub %u\n",
ring->idx, ring->name, ring->vm_inv_eng,
ring->funcs->vmhub);
}
/* Engine 17 is used for GART flushes */
for(i = 0; i < AMDGPU_MAX_VMHUBS; ++i)
BUG_ON(vm_inv_eng[i] > 17);
return amdgpu_irq_get(adev, &adev->gmc.vm_fault, 0);
}
static void gmc_v10_0_vram_gtt_location(struct amdgpu_device *adev,
struct amdgpu_gmc *mc)
{
u64 base = 0;
if (!amdgpu_sriov_vf(adev))
base = gfxhub_v2_0_get_fb_location(adev);
amdgpu_gmc_vram_location(adev, &adev->gmc, base);
amdgpu_gmc_gart_location(adev, mc);
/* base offset of vram pages */
adev->vm_manager.vram_base_offset = gfxhub_v2_0_get_mc_fb_offset(adev);
}
/**
* gmc_v10_0_mc_init - initialize the memory controller driver params
*
* @adev: amdgpu_device pointer
*
* Look up the amount of vram, vram width, and decide how to place
* vram and gart within the GPU's physical address space.
* Returns 0 for success.
*/
static int gmc_v10_0_mc_init(struct amdgpu_device *adev)
{
int chansize, numchan;
if (!amdgpu_emu_mode)
adev->gmc.vram_width = amdgpu_atomfirmware_get_vram_width(adev);
else {
/* hard code vram_width for emulation */
chansize = 128;
numchan = 1;
adev->gmc.vram_width = numchan * chansize;
}
/* Could aper size report 0 ? */
adev->gmc.aper_base = pci_resource_start(adev->pdev, 0);
adev->gmc.aper_size = pci_resource_len(adev->pdev, 0);
/* size in MB on si */
adev->gmc.mc_vram_size =
adev->nbio_funcs->get_memsize(adev) * 1024ULL * 1024ULL;
adev->gmc.real_vram_size = adev->gmc.mc_vram_size;
adev->gmc.visible_vram_size = adev->gmc.aper_size;
/* In case the PCI BAR is larger than the actual amount of vram */
if (adev->gmc.visible_vram_size > adev->gmc.real_vram_size)
adev->gmc.visible_vram_size = adev->gmc.real_vram_size;
/* set the gart size */
if (amdgpu_gart_size == -1) {
switch (adev->asic_type) {
case CHIP_NAVI10:
default:
adev->gmc.gart_size = 512ULL << 20;
break;
}
} else
adev->gmc.gart_size = (u64)amdgpu_gart_size << 20;
gmc_v10_0_vram_gtt_location(adev, &adev->gmc);
return 0;
}
static int gmc_v10_0_gart_init(struct amdgpu_device *adev)
{
int r;
if (adev->gart.bo) {
WARN(1, "NAVI10 PCIE GART already initialized\n");
return 0;
}
/* Initialize common gart structure */
r = amdgpu_gart_init(adev);
if (r)
return r;
adev->gart.table_size = adev->gart.num_gpu_pages * 8;
adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_NV10(MTYPE_UC) |
AMDGPU_PTE_EXECUTABLE;
return amdgpu_gart_table_vram_alloc(adev);
}
static unsigned gmc_v10_0_get_vbios_fb_size(struct amdgpu_device *adev)
{
u32 d1vga_control = RREG32_SOC15(DCE, 0, mmD1VGA_CONTROL);
unsigned size;
if (REG_GET_FIELD(d1vga_control, D1VGA_CONTROL, D1VGA_MODE_ENABLE)) {
size = 9 * 1024 * 1024; /* reserve 8MB for vga emulator and 1 MB for FB */
} else {
u32 viewport;
u32 pitch;
viewport = RREG32_SOC15(DCE, 0, mmHUBP0_DCSURF_PRI_VIEWPORT_DIMENSION);
pitch = RREG32_SOC15(DCE, 0, mmHUBPREQ0_DCSURF_SURFACE_PITCH);
size = (REG_GET_FIELD(viewport,
HUBP0_DCSURF_PRI_VIEWPORT_DIMENSION, PRI_VIEWPORT_HEIGHT) *
REG_GET_FIELD(pitch, HUBPREQ0_DCSURF_SURFACE_PITCH, PITCH) *
4);
}
/* return 0 if the pre-OS buffer uses up most of vram */
if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024)) {
DRM_ERROR("Warning: pre-OS buffer uses most of vram, \
be aware of gart table overwrite\n");
return 0;
}
return size;
}
static int gmc_v10_0_sw_init(void *handle)
{
int r;
int dma_bits;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
gfxhub_v2_0_init(adev);
mmhub_v2_0_init(adev);
spin_lock_init(&adev->gmc.invalidate_lock);
adev->gmc.vram_type = amdgpu_atomfirmware_get_vram_type(adev);
switch (adev->asic_type) {
case CHIP_NAVI10:
/*
* To fulfill 4-level page support,
* vm size is 256TB (48bit), maximum size of Navi10,
* block size 512 (9bit)
*/
amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
break;
default:
break;
}
/* This interrupt is VMC page fault.*/
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC,
VMC_1_0__SRCID__VM_FAULT,
&adev->gmc.vm_fault);
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UTCL2,
UTCL2_1_0__SRCID__FAULT,
&adev->gmc.vm_fault);
if (r)
return r;
/*
* Set the internal MC address mask This is the max address of the GPU's
* internal address space.
*/
adev->gmc.mc_mask = 0xffffffffffffULL; /* 48 bit MC */
/*
* Reserve 8M stolen memory for navi10 like vega10
* TODO: will check if it's really needed on asic.
*/
if (amdgpu_emu_mode == 1)
adev->gmc.stolen_size = 0;
else
adev->gmc.stolen_size = 9 * 1024 *1024;
/*
* Set DMA mask + need_dma32 flags.
* PCIE - can handle 44-bits.
* IGP - can handle 44-bits
* PCI - dma32 for legacy pci gart, 44 bits on navi10
*/
adev->need_dma32 = false;
dma_bits = adev->need_dma32 ? 32 : 44;
r = pci_set_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
if (r) {
adev->need_dma32 = true;
dma_bits = 32;
printk(KERN_WARNING "amdgpu: No suitable DMA available.\n");
}
r = pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(dma_bits));
if (r) {
pci_set_consistent_dma_mask(adev->pdev, DMA_BIT_MASK(32));
printk(KERN_WARNING "amdgpu: No coherent DMA available.\n");
}
r = gmc_v10_0_mc_init(adev);
if (r)
return r;
adev->gmc.stolen_size = gmc_v10_0_get_vbios_fb_size(adev);
/* Memory manager */
r = amdgpu_bo_init(adev);
if (r)
return r;
r = gmc_v10_0_gart_init(adev);
if (r)
return r;
/*
* number of VMs
* VMID 0 is reserved for System
* amdgpu graphics/compute will use VMIDs 1-7
* amdkfd will use VMIDs 8-15
*/
adev->vm_manager.id_mgr[AMDGPU_GFXHUB].num_ids = AMDGPU_NUM_OF_VMIDS;
adev->vm_manager.id_mgr[AMDGPU_MMHUB].num_ids = AMDGPU_NUM_OF_VMIDS;
amdgpu_vm_manager_init(adev);
return 0;
}
/**
* gmc_v8_0_gart_fini - vm fini callback
*
* @adev: amdgpu_device pointer
*
* Tears down the driver GART/VM setup (CIK).
*/
static void gmc_v10_0_gart_fini(struct amdgpu_device *adev)
{
amdgpu_gart_table_vram_free(adev);
amdgpu_gart_fini(adev);
}
static int gmc_v10_0_sw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
amdgpu_vm_manager_fini(adev);
gmc_v10_0_gart_fini(adev);
amdgpu_gem_force_release(adev);
amdgpu_bo_fini(adev);
return 0;
}
static void gmc_v10_0_init_golden_registers(struct amdgpu_device *adev)
{
switch (adev->asic_type) {
case CHIP_NAVI10:
break;
default:
break;
}
}
/**
* gmc_v10_0_gart_enable - gart enable
*
* @adev: amdgpu_device pointer
*/
static int gmc_v10_0_gart_enable(struct amdgpu_device *adev)
{
int r;
bool value;
u32 tmp;
if (adev->gart.bo == NULL) {
dev_err(adev->dev, "No VRAM object for PCIE GART.\n");
return -EINVAL;
}
r = amdgpu_gart_table_vram_pin(adev);
if (r)
return r;
r = gfxhub_v2_0_gart_enable(adev);
if (r)
return r;
r = mmhub_v2_0_gart_enable(adev);
if (r)
return r;
tmp = RREG32_SOC15(HDP, 0, mmHDP_MISC_CNTL);
tmp |= HDP_MISC_CNTL__FLUSH_INVALIDATE_CACHE_MASK;
WREG32_SOC15(HDP, 0, mmHDP_MISC_CNTL, tmp);
tmp = RREG32_SOC15(HDP, 0, mmHDP_HOST_PATH_CNTL);
WREG32_SOC15(HDP, 0, mmHDP_HOST_PATH_CNTL, tmp);
/* Flush HDP after it is initialized */
adev->nbio_funcs->hdp_flush(adev, NULL);
value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ?
false : true;
gfxhub_v2_0_set_fault_enable_default(adev, value);
mmhub_v2_0_set_fault_enable_default(adev, value);
gmc_v10_0_flush_gpu_tlb(adev, 0, 0);
DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
(unsigned)(adev->gmc.gart_size >> 20),
(unsigned long long)amdgpu_bo_gpu_offset(adev->gart.bo));
adev->gart.ready = true;
return 0;
}
static int gmc_v10_0_hw_init(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
/* The sequence of these two function calls matters.*/
gmc_v10_0_init_golden_registers(adev);
r = gmc_v10_0_gart_enable(adev);
if (r)
return r;
return 0;
}
/**
* gmc_v10_0_gart_disable - gart disable
*
* @adev: amdgpu_device pointer
*
* This disables all VM page table.
*/
static void gmc_v10_0_gart_disable(struct amdgpu_device *adev)
{
gfxhub_v2_0_gart_disable(adev);
mmhub_v2_0_gart_disable(adev);
amdgpu_gart_table_vram_unpin(adev);
}
static int gmc_v10_0_hw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (amdgpu_sriov_vf(adev)) {
/* full access mode, so don't touch any GMC register */
DRM_DEBUG("For SRIOV client, shouldn't do anything.\n");
return 0;
}
amdgpu_irq_put(adev, &adev->gmc.vm_fault, 0);
gmc_v10_0_gart_disable(adev);
return 0;
}
static int gmc_v10_0_suspend(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
gmc_v10_0_hw_fini(adev);
return 0;
}
static int gmc_v10_0_resume(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
r = gmc_v10_0_hw_init(adev);
if (r)
return r;
amdgpu_vmid_reset_all(adev);
return 0;
}
static bool gmc_v10_0_is_idle(void *handle)
{
/* MC is always ready in GMC v10.*/
return true;
}
static int gmc_v10_0_wait_for_idle(void *handle)
{
/* There is no need to wait for MC idle in GMC v10.*/
return 0;
}
static int gmc_v10_0_soft_reset(void *handle)
{
return 0;
}
static int gmc_v10_0_set_clockgating_state(void *handle,
enum amd_clockgating_state state)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
r = mmhub_v2_0_set_clockgating(adev, state);
if (r)
return r;
return athub_v2_0_set_clockgating(adev, state);
}
static void gmc_v10_0_get_clockgating_state(void *handle, u32 *flags)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
mmhub_v2_0_get_clockgating(adev, flags);
athub_v2_0_get_clockgating(adev, flags);
}
static int gmc_v10_0_set_powergating_state(void *handle,
enum amd_powergating_state state)
{
return 0;
}
const struct amd_ip_funcs gmc_v10_0_ip_funcs = {
.name = "gmc_v10_0",
.early_init = gmc_v10_0_early_init,
.late_init = gmc_v10_0_late_init,
.sw_init = gmc_v10_0_sw_init,
.sw_fini = gmc_v10_0_sw_fini,
.hw_init = gmc_v10_0_hw_init,
.hw_fini = gmc_v10_0_hw_fini,
.suspend = gmc_v10_0_suspend,
.resume = gmc_v10_0_resume,
.is_idle = gmc_v10_0_is_idle,
.wait_for_idle = gmc_v10_0_wait_for_idle,
.soft_reset = gmc_v10_0_soft_reset,
.set_clockgating_state = gmc_v10_0_set_clockgating_state,
.set_powergating_state = gmc_v10_0_set_powergating_state,
.get_clockgating_state = gmc_v10_0_get_clockgating_state,
};
const struct amdgpu_ip_block_version gmc_v10_0_ip_block =
{
.type = AMD_IP_BLOCK_TYPE_GMC,
.major = 10,
.minor = 0,
.rev = 0,
.funcs = &gmc_v10_0_ip_funcs,
};

30
drivers/gpu/drm/amd/amdgpu/gmc_v10_0.h Normal file
Просмотреть файл

@ -0,0 +1,30 @@
/*
* Copyright 2019 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 __GMC_V10_0_H__
#define __GMC_V10_0_H__
extern const struct amd_ip_funcs gmc_v10_0_ip_funcs;
extern const struct amdgpu_ip_block_version gmc_v10_0_ip_block;
#endif

19
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
Просмотреть файл

@ -535,22 +535,22 @@ static uint64_t gmc_v9_0_get_vm_pte_flags(struct amdgpu_device *adev,
switch (flags & AMDGPU_VM_MTYPE_MASK) {
case AMDGPU_VM_MTYPE_DEFAULT:
pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
pte_flag |= AMDGPU_PTE_MTYPE_VG10(MTYPE_NC);
break;
case AMDGPU_VM_MTYPE_NC:
pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
pte_flag |= AMDGPU_PTE_MTYPE_VG10(MTYPE_NC);
break;
case AMDGPU_VM_MTYPE_WC:
pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_WC);
pte_flag |= AMDGPU_PTE_MTYPE_VG10(MTYPE_WC);
break;
case AMDGPU_VM_MTYPE_CC:
pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_CC);
pte_flag |= AMDGPU_PTE_MTYPE_VG10(MTYPE_CC);
break;
case AMDGPU_VM_MTYPE_UC:
pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_UC);
pte_flag |= AMDGPU_PTE_MTYPE_VG10(MTYPE_UC);
break;
default:
pte_flag |= AMDGPU_PTE_MTYPE(MTYPE_NC);
pte_flag |= AMDGPU_PTE_MTYPE_VG10(MTYPE_NC);
break;
}
@ -733,9 +733,7 @@ static int gmc_v9_0_ecc_late_init(void *handle)
if (r)
goto interrupt;
r = amdgpu_ras_debugfs_create(adev, &fs_info);
if (r)
goto debugfs;
amdgpu_ras_debugfs_create(adev, &fs_info);
r = amdgpu_ras_sysfs_create(adev, &fs_info);
if (r)
@ -750,7 +748,6 @@ irq:
amdgpu_ras_sysfs_remove(adev, *ras_if);
sysfs:
amdgpu_ras_debugfs_remove(adev, *ras_if);
debugfs:
amdgpu_ras_interrupt_remove_handler(adev, &ih_info);
interrupt:
amdgpu_ras_feature_enable(adev, *ras_if, 0);
@ -919,7 +916,7 @@ static int gmc_v9_0_gart_init(struct amdgpu_device *adev)
if (r)
return r;
adev->gart.table_size = adev->gart.num_gpu_pages * 8;
adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE(MTYPE_UC) |
adev->gart.gart_pte_flags = AMDGPU_PTE_MTYPE_VG10(MTYPE_UC) |
AMDGPU_PTE_EXECUTABLE;
return amdgpu_gart_table_vram_alloc(adev);
}

366
drivers/gpu/drm/amd/amdgpu/mes_v10_1.c Normal file
Просмотреть файл

@ -0,0 +1,366 @@
/*
* Copyright 2019 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.
*
*/
#include <linux/firmware.h>
#include <linux/module.h>
#include "amdgpu.h"
#include "soc15_common.h"
#include "nv.h"
#include "gc/gc_10_1_0_offset.h"
#include "gc/gc_10_1_0_sh_mask.h"
MODULE_FIRMWARE("amdgpu/navi10_mes.bin");
static int mes_v10_1_add_hw_queue(struct amdgpu_mes *mes,
struct mes_add_queue_input *input)
{
return 0;
}
static int mes_v10_1_remove_hw_queue(struct amdgpu_mes *mes,
struct mes_remove_queue_input *input)
{
return 0;
}
static int mes_v10_1_suspend_gang(struct amdgpu_mes *mes,
struct mes_suspend_gang_input *input)
{
return 0;
}
static int mes_v10_1_resume_gang(struct amdgpu_mes *mes,
struct mes_resume_gang_input *input)
{
return 0;
}
static const struct amdgpu_mes_funcs mes_v10_1_funcs = {
.add_hw_queue = mes_v10_1_add_hw_queue,
.remove_hw_queue = mes_v10_1_remove_hw_queue,
.suspend_gang = mes_v10_1_suspend_gang,
.resume_gang = mes_v10_1_resume_gang,
};
static int mes_v10_1_init_microcode(struct amdgpu_device *adev)
{
const char *chip_name;
char fw_name[30];
int err;
const struct mes_firmware_header_v1_0 *mes_hdr;
switch (adev->asic_type) {
case CHIP_NAVI10:
chip_name = "navi10";
break;
default:
BUG();
}
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_mes.bin", chip_name);
err = request_firmware(&adev->mes.fw, fw_name, adev->dev);
if (err)
return err;
err = amdgpu_ucode_validate(adev->mes.fw);
if (err) {
release_firmware(adev->mes.fw);
adev->mes.fw = NULL;
return err;
}
mes_hdr = (const struct mes_firmware_header_v1_0 *)adev->mes.fw->data;
adev->mes.ucode_fw_version = le32_to_cpu(mes_hdr->mes_ucode_version);
adev->mes.ucode_fw_version =
le32_to_cpu(mes_hdr->mes_ucode_data_version);
adev->mes.uc_start_addr =
le32_to_cpu(mes_hdr->mes_uc_start_addr_lo) |
((uint64_t)(le32_to_cpu(mes_hdr->mes_uc_start_addr_hi)) << 32);
adev->mes.data_start_addr =
le32_to_cpu(mes_hdr->mes_data_start_addr_lo) |
((uint64_t)(le32_to_cpu(mes_hdr->mes_data_start_addr_hi)) << 32);
return 0;
}
static void mes_v10_1_free_microcode(struct amdgpu_device *adev)
{
release_firmware(adev->mes.fw);
adev->mes.fw = NULL;
}
static int mes_v10_1_allocate_ucode_buffer(struct amdgpu_device *adev)
{
int r;
const struct mes_firmware_header_v1_0 *mes_hdr;
const __le32 *fw_data;
unsigned fw_size;
mes_hdr = (const struct mes_firmware_header_v1_0 *)
adev->mes.fw->data;
fw_data = (const __le32 *)(adev->mes.fw->data +
le32_to_cpu(mes_hdr->mes_ucode_offset_bytes));
fw_size = le32_to_cpu(mes_hdr->mes_ucode_size_bytes);
r = amdgpu_bo_create_reserved(adev, fw_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
&adev->mes.ucode_fw_obj,
&adev->mes.ucode_fw_gpu_addr,
(void **)&adev->mes.ucode_fw_ptr);
if (r) {
dev_err(adev->dev, "(%d) failed to create mes fw bo\n", r);
return r;
}
memcpy(adev->mes.ucode_fw_ptr, fw_data, fw_size);
amdgpu_bo_kunmap(adev->mes.ucode_fw_obj);
amdgpu_bo_unreserve(adev->mes.ucode_fw_obj);
return 0;
}
static int mes_v10_1_allocate_ucode_data_buffer(struct amdgpu_device *adev)
{
int r;
const struct mes_firmware_header_v1_0 *mes_hdr;
const __le32 *fw_data;
unsigned fw_size;
mes_hdr = (const struct mes_firmware_header_v1_0 *)
adev->mes.fw->data;
fw_data = (const __le32 *)(adev->mes.fw->data +
le32_to_cpu(mes_hdr->mes_ucode_data_offset_bytes));
fw_size = le32_to_cpu(mes_hdr->mes_ucode_data_size_bytes);
r = amdgpu_bo_create_reserved(adev, fw_size,
64 * 1024, AMDGPU_GEM_DOMAIN_GTT,
&adev->mes.data_fw_obj,
&adev->mes.data_fw_gpu_addr,
(void **)&adev->mes.data_fw_ptr);
if (r) {
dev_err(adev->dev, "(%d) failed to create mes data fw bo\n", r);
return r;
}
memcpy(adev->mes.data_fw_ptr, fw_data, fw_size);
amdgpu_bo_kunmap(adev->mes.data_fw_obj);
amdgpu_bo_unreserve(adev->mes.data_fw_obj);
return 0;
}
static void mes_v10_1_free_ucode_buffers(struct amdgpu_device *adev)
{
amdgpu_bo_free_kernel(&adev->mes.data_fw_obj,
&adev->mes.data_fw_gpu_addr,
(void **)&adev->mes.data_fw_ptr);
amdgpu_bo_free_kernel(&adev->mes.ucode_fw_obj,
&adev->mes.ucode_fw_gpu_addr,
(void **)&adev->mes.ucode_fw_ptr);
}
static void mes_v10_1_enable(struct amdgpu_device *adev, bool enable)
{
uint32_t data = 0;
if (enable) {
data = RREG32_SOC15(GC, 0, mmCP_MES_CNTL);
data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_RESET, 1);
WREG32_SOC15(GC, 0, mmCP_MES_CNTL, data);
/* set ucode start address */
WREG32_SOC15(GC, 0, mmCP_MES_PRGRM_CNTR_START,
(uint32_t)(adev->mes.uc_start_addr) >> 2);
/* clear BYPASS_UNCACHED to avoid hangs after interrupt. */
data = RREG32_SOC15(GC, 0, mmCP_MES_DC_OP_CNTL);
data = REG_SET_FIELD(data, CP_MES_DC_OP_CNTL,
BYPASS_UNCACHED, 0);
WREG32_SOC15(GC, 0, mmCP_MES_DC_OP_CNTL, data);
/* unhalt MES and activate pipe0 */
data = REG_SET_FIELD(0, CP_MES_CNTL, MES_PIPE0_ACTIVE, 1);
WREG32_SOC15(GC, 0, mmCP_MES_CNTL, data);
} else {
data = RREG32_SOC15(GC, 0, mmCP_MES_CNTL);
data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_ACTIVE, 0);
data = REG_SET_FIELD(data, CP_MES_CNTL,
MES_INVALIDATE_ICACHE, 1);
data = REG_SET_FIELD(data, CP_MES_CNTL, MES_PIPE0_RESET, 1);
data = REG_SET_FIELD(data, CP_MES_CNTL, MES_HALT, 1);
WREG32_SOC15(GC, 0, mmCP_MES_CNTL, data);
}
}
/* This function is for backdoor MES firmware */
static int mes_v10_1_load_microcode(struct amdgpu_device *adev)
{
int r;
uint32_t data;
if (!adev->mes.fw)
return -EINVAL;
r = mes_v10_1_allocate_ucode_buffer(adev);
if (r)
return r;
r = mes_v10_1_allocate_ucode_data_buffer(adev);
if (r) {
mes_v10_1_free_ucode_buffers(adev);
return r;
}
mes_v10_1_enable(adev, false);
WREG32_SOC15(GC, 0, mmCP_MES_IC_BASE_CNTL, 0);
mutex_lock(&adev->srbm_mutex);
/* me=3, pipe=0, queue=0 */
nv_grbm_select(adev, 3, 0, 0, 0);
/* set ucode start address */
WREG32_SOC15(GC, 0, mmCP_MES_PRGRM_CNTR_START,
(uint32_t)(adev->mes.uc_start_addr) >> 2);
/* set ucode fimrware address */
WREG32_SOC15(GC, 0, mmCP_MES_IC_BASE_LO,
lower_32_bits(adev->mes.ucode_fw_gpu_addr));
WREG32_SOC15(GC, 0, mmCP_MES_IC_BASE_HI,
upper_32_bits(adev->mes.ucode_fw_gpu_addr));
/* set ucode instruction cache boundary to 2M-1 */
WREG32_SOC15(GC, 0, mmCP_MES_MIBOUND_LO, 0x1FFFFF);
/* set ucode data firmware address */
WREG32_SOC15(GC, 0, mmCP_MES_MDBASE_LO,
lower_32_bits(adev->mes.data_fw_gpu_addr));
WREG32_SOC15(GC, 0, mmCP_MES_MDBASE_HI,
upper_32_bits(adev->mes.data_fw_gpu_addr));
/* Set 0x3FFFF (256K-1) to CP_MES_MDBOUND_LO */
WREG32_SOC15(GC, 0, mmCP_MES_MDBOUND_LO, 0x3FFFF);
/* invalidate ICACHE */
data = RREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL);
data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 0);
data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, INVALIDATE_CACHE, 1);
WREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL, data);
/* prime the ICACHE. */
data = RREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL);
data = REG_SET_FIELD(data, CP_MES_IC_OP_CNTL, PRIME_ICACHE, 1);
WREG32_SOC15(GC, 0, mmCP_MES_IC_OP_CNTL, data);
nv_grbm_select(adev, 0, 0, 0, 0);
mutex_unlock(&adev->srbm_mutex);
return 0;
}
static int mes_v10_1_sw_init(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
r = mes_v10_1_init_microcode(adev);
if (r)
return r;
return 0;
}
static int mes_v10_1_sw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
mes_v10_1_free_microcode(adev);
return 0;
}
static int mes_v10_1_hw_init(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) {
r = mes_v10_1_load_microcode(adev);
if (r) {
DRM_ERROR("failed to MES fw, r=%d\n", r);
return r;
}
} else {
DRM_ERROR("only support direct fw loading on MES\n");
return -EINVAL;
}
mes_v10_1_enable(adev, true);
return 0;
}
static int mes_v10_1_hw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
mes_v10_1_enable(adev, false);
if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)
mes_v10_1_free_ucode_buffers(adev);
return 0;
}
static int mes_v10_1_suspend(void *handle)
{
return 0;
}
static int mes_v10_1_resume(void *handle)
{
return 0;
}
static const struct amd_ip_funcs mes_v10_1_ip_funcs = {
.name = "mes_v10_1",
.sw_init = mes_v10_1_sw_init,
.sw_fini = mes_v10_1_sw_fini,
.hw_init = mes_v10_1_hw_init,
.hw_fini = mes_v10_1_hw_fini,
.suspend = mes_v10_1_suspend,
.resume = mes_v10_1_resume,
};
const struct amdgpu_ip_block_version mes_v10_1_ip_block = {
.type = AMD_IP_BLOCK_TYPE_MES,
.major = 10,
.minor = 1,
.rev = 0,
.funcs = &mes_v10_1_ip_funcs,
};

29
drivers/gpu/drm/amd/amdgpu/mes_v10_1.h Normal file
Просмотреть файл

@ -0,0 +1,29 @@
/*
* Copyright 2019 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 __MES_V10_1_H__
#define __MES_v10_1_H__
extern const struct amdgpu_ip_block_version mes_v10_1_ip_block;
#endif

444
drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c Normal file
Просмотреть файл

@ -0,0 +1,444 @@
/*
* Copyright 2019 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.
*
*/
#include "amdgpu.h"
#include "mmhub_v2_0.h"
#include "mmhub/mmhub_2_0_0_offset.h"
#include "mmhub/mmhub_2_0_0_sh_mask.h"
#include "mmhub/mmhub_2_0_0_default.h"
#include "navi10_enum.h"
#include "soc15_common.h"
static void mmhub_v2_0_init_gart_pt_regs(struct amdgpu_device *adev)
{
uint64_t value = amdgpu_gmc_pd_addr(adev->gart.bo);
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32,
lower_32_bits(value));
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32,
upper_32_bits(value));
}
static void mmhub_v2_0_init_gart_aperture_regs(struct amdgpu_device *adev)
{
mmhub_v2_0_init_gart_pt_regs(adev);
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
(u32)(adev->gmc.gart_start >> 12));
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
(u32)(adev->gmc.gart_start >> 44));
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
(u32)(adev->gmc.gart_end >> 12));
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
(u32)(adev->gmc.gart_end >> 44));
}
static void mmhub_v2_0_init_system_aperture_regs(struct amdgpu_device *adev)
{
uint64_t value;
uint32_t tmp;
/* Disable AGP. */
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BASE, 0);
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_TOP, 0);
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_AGP_BOT, 0x00FFFFFF);
/* Program the system aperture low logical page number. */
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_LOW_ADDR,
adev->gmc.vram_start >> 18);
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_HIGH_ADDR,
adev->gmc.vram_end >> 18);
/* Set default page address. */
value = adev->vram_scratch.gpu_addr - adev->gmc.vram_start +
adev->vm_manager.vram_base_offset;
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_LSB,
(u32)(value >> 12));
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR_MSB,
(u32)(value >> 44));
/* Program "protection fault". */
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_LO32,
(u32)(adev->dummy_page_addr >> 12));
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_DEFAULT_ADDR_HI32,
(u32)((u64)adev->dummy_page_addr >> 44));
tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL2,
ACTIVE_PAGE_MIGRATION_PTE_READ_RETRY, 1);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL2, tmp);
}
static void mmhub_v2_0_init_tlb_regs(struct amdgpu_device *adev)
{
uint32_t tmp;
/* Setup TLB control */
tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 1);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, SYSTEM_ACCESS_MODE, 3);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
ENABLE_ADVANCED_DRIVER_MODEL, 1);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
SYSTEM_APERTURE_UNMAPPED_ACCESS, 0);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ECO_BITS, 0);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
MTYPE, MTYPE_UC); /* UC, uncached */
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp);
}
static void mmhub_v2_0_init_cache_regs(struct amdgpu_device *adev)
{
uint32_t tmp;
/* Setup L2 cache */
tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 1);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL,
ENABLE_DEFAULT_PAGE_OUT_TO_SYSTEM_MEMORY, 1);
/* XXX for emulation, Refer to closed source code.*/
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
0);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, PDE_FAULT_CLASSIFICATION, 1);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, CONTEXT1_IDENTITY_ACCESS_MODE, 1);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, IDENTITY_MODE_FRAGMENT_SIZE, 0);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp);
tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_ALL_L1_TLBS, 1);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL2, INVALIDATE_L2_CACHE, 1);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL2, tmp);
tmp = mmMMVM_L2_CNTL3_DEFAULT;
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, tmp);
tmp = mmMMVM_L2_CNTL4_DEFAULT;
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PDE_REQUEST_PHYSICAL, 0);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL4, VMC_TAP_PTE_REQUEST_PHYSICAL, 0);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL4, tmp);
}
static void mmhub_v2_0_enable_system_domain(struct amdgpu_device *adev)
{
uint32_t tmp;
tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, ENABLE_CONTEXT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT0_CNTL, PAGE_TABLE_DEPTH, 0);
WREG32_SOC15(MMHUB, 0, mmMMVM_CONTEXT0_CNTL, tmp);
}
static void mmhub_v2_0_disable_identity_aperture(struct amdgpu_device *adev)
{
WREG32_SOC15(MMHUB, 0,
mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_LO32,
0xFFFFFFFF);
WREG32_SOC15(MMHUB, 0,
mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_LOW_ADDR_HI32,
0x0000000F);
WREG32_SOC15(MMHUB, 0,
mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_LO32, 0);
WREG32_SOC15(MMHUB, 0,
mmMMVM_L2_CONTEXT1_IDENTITY_APERTURE_HIGH_ADDR_HI32, 0);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_LO32,
0);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CONTEXT_IDENTITY_PHYSICAL_OFFSET_HI32,
0);
}
static void mmhub_v2_0_setup_vmid_config(struct amdgpu_device *adev)
{
int i;
uint32_t tmp;
for (i = 0; i <= 14; i++) {
tmp = RREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, i);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, ENABLE_CONTEXT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL, PAGE_TABLE_DEPTH,
adev->vm_manager.num_level);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT,
1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
VALID_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
READ_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
PAGE_TABLE_BLOCK_SIZE,
adev->vm_manager.block_size - 9);
/* Send no-retry XNACK on fault to suppress VM fault storm. */
tmp = REG_SET_FIELD(tmp, MMVM_CONTEXT1_CNTL,
RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_CNTL, i, tmp);
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_LO32, i*2,
lower_32_bits(adev->vm_manager.max_pfn - 1));
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT1_PAGE_TABLE_END_ADDR_HI32, i*2,
upper_32_bits(adev->vm_manager.max_pfn - 1));
}
}
static void mmhub_v2_0_program_invalidation(struct amdgpu_device *adev)
{
unsigned i;
for (i = 0; i < 18; ++i) {
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_LO32,
2 * i, 0xffffffff);
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ADDR_RANGE_HI32,
2 * i, 0x1f);
}
}
int mmhub_v2_0_gart_enable(struct amdgpu_device *adev)
{
if (amdgpu_sriov_vf(adev)) {
/*
* MMMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are
* VF copy registers so vbios post doesn't program them, for
* SRIOV driver need to program them
*/
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_FB_LOCATION_BASE,
adev->gmc.vram_start >> 24);
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_FB_LOCATION_TOP,
adev->gmc.vram_end >> 24);
}
/* GART Enable. */
mmhub_v2_0_init_gart_aperture_regs(adev);
mmhub_v2_0_init_system_aperture_regs(adev);
mmhub_v2_0_init_tlb_regs(adev);
mmhub_v2_0_init_cache_regs(adev);
mmhub_v2_0_enable_system_domain(adev);
mmhub_v2_0_disable_identity_aperture(adev);
mmhub_v2_0_setup_vmid_config(adev);
mmhub_v2_0_program_invalidation(adev);
return 0;
}
void mmhub_v2_0_gart_disable(struct amdgpu_device *adev)
{
u32 tmp;
u32 i;
/* Disable all tables */
for (i = 0; i < 16; i++)
WREG32_SOC15_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL, i, 0);
/* Setup TLB control */
tmp = RREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL, ENABLE_L1_TLB, 0);
tmp = REG_SET_FIELD(tmp, MMMC_VM_MX_L1_TLB_CNTL,
ENABLE_ADVANCED_DRIVER_MODEL, 0);
WREG32_SOC15(MMHUB, 0, mmMMMC_VM_MX_L1_TLB_CNTL, tmp);
/* Setup L2 cache */
tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL);
tmp = REG_SET_FIELD(tmp, MMVM_L2_CNTL, ENABLE_L2_CACHE, 0);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL, tmp);
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_CNTL3, 0);
}
/**
* mmhub_v2_0_set_fault_enable_default - update GART/VM fault handling
*
* @adev: amdgpu_device pointer
* @value: true redirects VM faults to the default page
*/
void mmhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev, bool value)
{
u32 tmp;
tmp = RREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
RANGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
PDE0_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
PDE1_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
PDE2_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
TRANSLATE_FURTHER_PROTECTION_FAULT_ENABLE_DEFAULT,
value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
NACK_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
DUMMY_PAGE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
VALID_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
READ_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
WRITE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
EXECUTE_PROTECTION_FAULT_ENABLE_DEFAULT, value);
if (!value) {
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
CRASH_ON_NO_RETRY_FAULT, 1);
tmp = REG_SET_FIELD(tmp, MMVM_L2_PROTECTION_FAULT_CNTL,
CRASH_ON_RETRY_FAULT, 1);
}
WREG32_SOC15(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL, tmp);
}
void mmhub_v2_0_init(struct amdgpu_device *adev)
{
struct amdgpu_vmhub *hub = &adev->vmhub[AMDGPU_MMHUB];
hub->ctx0_ptb_addr_lo32 =
SOC15_REG_OFFSET(MMHUB, 0,
mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_LO32);
hub->ctx0_ptb_addr_hi32 =
SOC15_REG_OFFSET(MMHUB, 0,
mmMMVM_CONTEXT0_PAGE_TABLE_BASE_ADDR_HI32);
hub->vm_inv_eng0_req =
SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_REQ);
hub->vm_inv_eng0_ack =
SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_INVALIDATE_ENG0_ACK);
hub->vm_context0_cntl =
SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_CONTEXT0_CNTL);
hub->vm_l2_pro_fault_status =
SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_STATUS);
hub->vm_l2_pro_fault_cntl =
SOC15_REG_OFFSET(MMHUB, 0, mmMMVM_L2_PROTECTION_FAULT_CNTL);
}
static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *adev,
bool enable)
{
uint32_t def, data, def1, data1;
def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG);
def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_MGCG)) {
data |= MM_ATC_L2_MISC_CG__ENABLE_MASK;
data1 &= ~(DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
} else {
data &= ~MM_ATC_L2_MISC_CG__ENABLE_MASK;
data1 |= (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK);
}
if (def != data)
WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data);
if (def1 != data1)
WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2, data1);
}
static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *adev,
bool enable)
{
uint32_t def, data;
def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG);
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_MC_LS))
data |= MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
else
data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK;
if (def != data)
WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data);
}
int mmhub_v2_0_set_clockgating(struct amdgpu_device *adev,
enum amd_clockgating_state state)
{
if (amdgpu_sriov_vf(adev))
return 0;
switch (adev->asic_type) {
case CHIP_NAVI10:
mmhub_v2_0_update_medium_grain_clock_gating(adev,
state == AMD_CG_STATE_GATE ? true : false);
mmhub_v2_0_update_medium_grain_light_sleep(adev,
state == AMD_CG_STATE_GATE ? true : false);
break;
default:
break;
}
return 0;
}
void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags)
{
int data, data1;
if (amdgpu_sriov_vf(adev))
*flags = 0;
data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG);
data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2);
/* AMD_CG_SUPPORT_MC_MGCG */
if ((data & MM_ATC_L2_MISC_CG__ENABLE_MASK) &&
!(data1 & (DAGB0_CNTL_MISC2__DISABLE_WRREQ_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_WRRET_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_RDREQ_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_RDRET_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_TLBWR_CG_MASK |
DAGB0_CNTL_MISC2__DISABLE_TLBRD_CG_MASK)))
*flags |= AMD_CG_SUPPORT_MC_MGCG;
/* AMD_CG_SUPPORT_MC_LS */
if (data & MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK)
*flags |= AMD_CG_SUPPORT_MC_LS;
}

35
drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.h Normal file
Просмотреть файл

@ -0,0 +1,35 @@
/*
* Copyright 2019 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 __MMHUB_V2_0_H__
#define __MMHUB_V2_0_H__
int mmhub_v2_0_gart_enable(struct amdgpu_device *adev);
void mmhub_v2_0_gart_disable(struct amdgpu_device *adev);
void mmhub_v2_0_set_fault_enable_default(struct amdgpu_device *adev,
bool value);
void mmhub_v2_0_init(struct amdgpu_device *adev);
int mmhub_v2_0_set_clockgating(struct amdgpu_device *adev,
enum amd_clockgating_state state);
void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u32 *flags);
#endif

15
drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
Просмотреть файл

@ -451,19 +451,16 @@ void xgpu_ai_mailbox_put_irq(struct amdgpu_device *adev)
static void xgpu_ai_init_reg_access_mode(struct amdgpu_device *adev)
{
uint32_t rlc_fw_ver = RREG32_SOC15(GC, 0, mmRLC_GPM_GENERAL_6);
uint32_t sos_fw_ver = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_58);
adev->virt.reg_access_mode = AMDGPU_VIRT_REG_ACCESS_LEGACY;
if (rlc_fw_ver >= 0x5d)
adev->virt.reg_access_mode |= AMDGPU_VIRT_REG_ACCESS_RLC;
/* Enable L1 security reg access mode by defaul, as non-security VF
* will no longer be supported.
*/
adev->virt.reg_access_mode |= AMDGPU_VIRT_REG_ACCESS_RLC;
if (sos_fw_ver >= 0x80455)
adev->virt.reg_access_mode |= AMDGPU_VIRT_REG_ACCESS_PSP_PRG_IH;
adev->virt.reg_access_mode |= AMDGPU_VIRT_REG_ACCESS_PSP_PRG_IH;
if (sos_fw_ver >= 0x8045b)
adev->virt.reg_access_mode |= AMDGPU_VIRT_REG_SKIP_SEETING;
adev->virt.reg_access_mode |= AMDGPU_VIRT_REG_SKIP_SEETING;
}
const struct amdgpu_virt_ops xgpu_ai_virt_ops = {

486
drivers/gpu/drm/amd/amdgpu/navi10_ih.c Normal file
Просмотреть файл

@ -0,0 +1,486 @@
/*
* Copyright 2019 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.
*
*/
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_ih.h"
#include "oss/osssys_5_0_0_offset.h"
#include "oss/osssys_5_0_0_sh_mask.h"
#include "soc15_common.h"
#include "navi10_ih.h"
static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev);
/**
* navi10_ih_enable_interrupts - Enable the interrupt ring buffer
*
* @adev: amdgpu_device pointer
*
* Enable the interrupt ring buffer (NAVI10).
*/
static void navi10_ih_enable_interrupts(struct amdgpu_device *adev)
{
u32 ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
adev->irq.ih.enabled = true;
}
/**
* navi10_ih_disable_interrupts - Disable the interrupt ring buffer
*
* @adev: amdgpu_device pointer
*
* Disable the interrupt ring buffer (NAVI10).
*/
static void navi10_ih_disable_interrupts(struct amdgpu_device *adev)
{
u32 ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_ENABLE, 0);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
adev->irq.ih.enabled = false;
adev->irq.ih.rptr = 0;
}
static uint32_t navi10_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
{
int rb_bufsz = order_base_2(ih->ring_size / 4);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
MC_SPACE, ih->use_bus_addr ? 1 : 4);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
WPTR_OVERFLOW_CLEAR, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
WPTR_OVERFLOW_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
* value is written to memory
*/
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
WPTR_WRITEBACK_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
return ih_rb_cntl;
}
/**
* navi10_ih_irq_init - init and enable the interrupt ring
*
* @adev: amdgpu_device pointer
*
* Allocate a ring buffer for the interrupt controller,
* enable the RLC, disable interrupts, enable the IH
* ring buffer and enable it (NAVI).
* Called at device load and reume.
* Returns 0 for success, errors for failure.
*/
static int navi10_ih_irq_init(struct amdgpu_device *adev)
{
struct amdgpu_ih_ring *ih = &adev->irq.ih;
int ret = 0;
u32 ih_rb_cntl, ih_doorbell_rtpr, ih_chicken;
u32 tmp;
/* disable irqs */
navi10_ih_disable_interrupts(adev);
adev->nbio_funcs->ih_control(adev);
/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, ih->gpu_addr >> 8);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, (ih->gpu_addr >> 40) & 0xff);
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
ih_rb_cntl = navi10_ih_rb_cntl(ih, ih_rb_cntl);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM,
!!adev->irq.msi_enabled);
if (unlikely(adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT)) {
if (ih->use_bus_addr) {
ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN);
ih_chicken = REG_SET_FIELD(ih_chicken,
IH_CHICKEN, MC_SPACE_GPA_ENABLE, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN, ih_chicken);
}
}
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
/* set the writeback address whether it's enabled or not */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO,
lower_32_bits(ih->wptr_addr));
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI,
upper_32_bits(ih->wptr_addr) & 0xFFFF);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
ih_doorbell_rtpr = RREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR);
if (ih->use_doorbell) {
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
IH_DOORBELL_RPTR, OFFSET,
ih->doorbell_index);
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
IH_DOORBELL_RPTR, ENABLE, 1);
} else {
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
IH_DOORBELL_RPTR, ENABLE, 0);
}
WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR, ih_doorbell_rtpr);
adev->nbio_funcs->ih_doorbell_range(adev, ih->use_doorbell,
ih->doorbell_index);
tmp = RREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL);
tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
CLIENT18_IS_STORM_CLIENT, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL, tmp);
tmp = RREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL);
tmp = REG_SET_FIELD(tmp, IH_INT_FLOOD_CNTL, FLOOD_CNTL_ENABLE, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_INT_FLOOD_CNTL, tmp);
pci_set_master(adev->pdev);
/* enable interrupts */
navi10_ih_enable_interrupts(adev);
return ret;
}
/**
* navi10_ih_irq_disable - disable interrupts
*
* @adev: amdgpu_device pointer
*
* Disable interrupts on the hw (NAVI10).
*/
static void navi10_ih_irq_disable(struct amdgpu_device *adev)
{
navi10_ih_disable_interrupts(adev);
/* Wait and acknowledge irq */
mdelay(1);
}
/**
* navi10_ih_get_wptr - get the IH ring buffer wptr
*
* @adev: amdgpu_device pointer
*
* Get the IH ring buffer wptr from either the register
* or the writeback memory buffer (NAVI10). Also check for
* ring buffer overflow and deal with it.
* Returns the value of the wptr.
*/
static u32 navi10_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, reg, tmp;
wptr = le32_to_cpu(*ih->wptr_cpu);
if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
goto out;
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR);
wptr = RREG32_NO_KIQ(reg);
if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
goto out;
wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
/* When a ring buffer overflow happen start parsing interrupt
* from the last not overwritten vector (wptr + 32). Hopefully
* this should allow us to catch up.
*/
tmp = (wptr + 32) & ih->ptr_mask;
dev_warn(adev->dev, "IH ring buffer overflow "
"(0x%08X, 0x%08X, 0x%08X)\n",
wptr, ih->rptr, tmp);
ih->rptr = tmp;
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL);
tmp = RREG32_NO_KIQ(reg);
tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
WREG32_NO_KIQ(reg, tmp);
out:
return (wptr & ih->ptr_mask);
}
/**
* navi10_ih_decode_iv - decode an interrupt vector
*
* @adev: amdgpu_device pointer
*
* Decodes the interrupt vector at the current rptr
* position and also advance the position.
*/
static void navi10_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
/* wptr/rptr are in bytes! */
u32 ring_index = ih->rptr >> 2;
uint32_t dw[8];
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
dw[4] = le32_to_cpu(ih->ring[ring_index + 4]);
dw[5] = le32_to_cpu(ih->ring[ring_index + 5]);
dw[6] = le32_to_cpu(ih->ring[ring_index + 6]);
dw[7] = le32_to_cpu(ih->ring[ring_index + 7]);
entry->client_id = dw[0] & 0xff;
entry->src_id = (dw[0] >> 8) & 0xff;
entry->ring_id = (dw[0] >> 16) & 0xff;
entry->vmid = (dw[0] >> 24) & 0xf;
entry->vmid_src = (dw[0] >> 31);
entry->timestamp = dw[1] | ((u64)(dw[2] & 0xffff) << 32);
entry->timestamp_src = dw[2] >> 31;
entry->pasid = dw[3] & 0xffff;
entry->pasid_src = dw[3] >> 31;
entry->src_data[0] = dw[4];
entry->src_data[1] = dw[5];
entry->src_data[2] = dw[6];
entry->src_data[3] = dw[7];
/* wptr/rptr are in bytes! */
ih->rptr += 32;
}
/**
* navi10_ih_set_rptr - set the IH ring buffer rptr
*
* @adev: amdgpu_device pointer
*
* Set the IH ring buffer rptr.
*/
static void navi10_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
if (ih->use_doorbell) {
/* XXX check if swapping is necessary on BE */
*ih->rptr_cpu = ih->rptr;
WDOORBELL32(ih->doorbell_index, ih->rptr);
} else
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, ih->rptr);
}
static int navi10_ih_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
navi10_ih_set_interrupt_funcs(adev);
return 0;
}
static int navi10_ih_sw_init(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
bool use_bus_addr;
/* use gpu virtual address for ih ring
* until ih_checken is programmed to allow
* use bus address for ih ring by psp bl */
use_bus_addr =
(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) ? false : true;
r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, use_bus_addr);
if (r)
return r;
adev->irq.ih.use_doorbell = true;
adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
r = amdgpu_irq_init(adev);
return r;
}
static int navi10_ih_sw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
amdgpu_irq_fini(adev);
amdgpu_ih_ring_fini(adev, &adev->irq.ih);
return 0;
}
static int navi10_ih_hw_init(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
r = navi10_ih_irq_init(adev);
if (r)
return r;
return 0;
}
static int navi10_ih_hw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
navi10_ih_irq_disable(adev);
return 0;
}
static int navi10_ih_suspend(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
return navi10_ih_hw_fini(adev);
}
static int navi10_ih_resume(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
return navi10_ih_hw_init(adev);
}
static bool navi10_ih_is_idle(void *handle)
{
/* todo */
return true;
}
static int navi10_ih_wait_for_idle(void *handle)
{
/* todo */
return -ETIMEDOUT;
}
static int navi10_ih_soft_reset(void *handle)
{
/* todo */
return 0;
}
static void navi10_ih_update_clockgating_state(struct amdgpu_device *adev,
bool enable)
{
uint32_t data, def, field_val;
if (adev->cg_flags & AMD_CG_SUPPORT_IH_CG) {
def = data = RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL);
field_val = enable ? 0 : 1;
data = REG_SET_FIELD(data, IH_CLK_CTRL,
DBUS_MUX_CLK_SOFT_OVERRIDE, field_val);
data = REG_SET_FIELD(data, IH_CLK_CTRL,
OSSSYS_SHARE_CLK_SOFT_OVERRIDE, field_val);
data = REG_SET_FIELD(data, IH_CLK_CTRL,
LIMIT_SMN_CLK_SOFT_OVERRIDE, field_val);
data = REG_SET_FIELD(data, IH_CLK_CTRL,
DYN_CLK_SOFT_OVERRIDE, field_val);
data = REG_SET_FIELD(data, IH_CLK_CTRL,
REG_CLK_SOFT_OVERRIDE, field_val);
if (def != data)
WREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL, data);
}
return;
}
static int navi10_ih_set_clockgating_state(void *handle,
enum amd_clockgating_state state)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
navi10_ih_update_clockgating_state(adev,
state == AMD_CG_STATE_GATE ? true : false);
return 0;
}
static int navi10_ih_set_powergating_state(void *handle,
enum amd_powergating_state state)
{
return 0;
}
static void navi10_ih_get_clockgating_state(void *handle, u32 *flags)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (!RREG32_SOC15(OSSSYS, 0, mmIH_CLK_CTRL))
*flags |= AMD_CG_SUPPORT_IH_CG;
return;
}
static const struct amd_ip_funcs navi10_ih_ip_funcs = {
.name = "navi10_ih",
.early_init = navi10_ih_early_init,
.late_init = NULL,
.sw_init = navi10_ih_sw_init,
.sw_fini = navi10_ih_sw_fini,
.hw_init = navi10_ih_hw_init,
.hw_fini = navi10_ih_hw_fini,
.suspend = navi10_ih_suspend,
.resume = navi10_ih_resume,
.is_idle = navi10_ih_is_idle,
.wait_for_idle = navi10_ih_wait_for_idle,
.soft_reset = navi10_ih_soft_reset,
.set_clockgating_state = navi10_ih_set_clockgating_state,
.set_powergating_state = navi10_ih_set_powergating_state,
.get_clockgating_state = navi10_ih_get_clockgating_state,
};
static const struct amdgpu_ih_funcs navi10_ih_funcs = {
.get_wptr = navi10_ih_get_wptr,
.decode_iv = navi10_ih_decode_iv,
.set_rptr = navi10_ih_set_rptr
};
static void navi10_ih_set_interrupt_funcs(struct amdgpu_device *adev)
{
if (adev->irq.ih_funcs == NULL)
adev->irq.ih_funcs = &navi10_ih_funcs;
}
const struct amdgpu_ip_block_version navi10_ih_ip_block =
{
.type = AMD_IP_BLOCK_TYPE_IH,
.major = 5,
.minor = 0,
.rev = 0,
.funcs = &navi10_ih_ip_funcs,
};

29
drivers/gpu/drm/amd/amdgpu/navi10_ih.h Normal file
Просмотреть файл

@ -0,0 +1,29 @@
/*
* Copyright 2019 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 __NAVI10_IH_H__
#define __NAVI10_IH_H__
extern const struct amdgpu_ip_block_version navi10_ih_ip_block;
#endif

68
drivers/gpu/drm/amd/amdgpu/navi10_reg_init.c Normal file
Просмотреть файл

@ -0,0 +1,68 @@
/*
* Copyright 2018 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.
*
*/
#include "amdgpu.h"
#include "nv.h"
#include "soc15_common.h"
#include "soc15_hw_ip.h"
#include "navi10_ip_offset.h"
int navi10_reg_base_init(struct amdgpu_device *adev)
{
int r, i;
if (amdgpu_discovery) {
r = amdgpu_discovery_reg_base_init(adev);
if (r) {
DRM_WARN("failed to init reg base from ip discovery table, "
"fallback to legacy init method\n");
goto legacy_init;
}
return 0;
}
legacy_init:
for (i = 0 ; i < MAX_INSTANCE ; ++i) {
adev->reg_offset[GC_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
adev->reg_offset[HDP_HWIP][i] = (uint32_t *)(&(HDP_BASE.instance[i]));
adev->reg_offset[MMHUB_HWIP][i] = (uint32_t *)(&(MMHUB_BASE.instance[i]));
adev->reg_offset[ATHUB_HWIP][i] = (uint32_t *)(&(ATHUB_BASE.instance[i]));
adev->reg_offset[NBIO_HWIP][i] = (uint32_t *)(&(NBIO_BASE.instance[i]));
adev->reg_offset[MP0_HWIP][i] = (uint32_t *)(&(MP0_BASE.instance[i]));
adev->reg_offset[MP1_HWIP][i] = (uint32_t *)(&(MP1_BASE.instance[i]));
adev->reg_offset[VCN_HWIP][i] = (uint32_t *)(&(VCN_BASE.instance[i]));
adev->reg_offset[DF_HWIP][i] = (uint32_t *)(&(DF_BASE.instance[i]));
adev->reg_offset[DCE_HWIP][i] = (uint32_t *)(&(DCN_BASE.instance[i]));
adev->reg_offset[OSSSYS_HWIP][i] = (uint32_t *)(&(OSSSYS_BASE.instance[i]));
adev->reg_offset[SDMA0_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
adev->reg_offset[SDMA1_HWIP][i] = (uint32_t *)(&(GC_BASE.instance[i]));
adev->reg_offset[SMUIO_HWIP][i] = (uint32_t *)(&(SMUIO_BASE.instance[i]));
adev->reg_offset[THM_HWIP][i] = (uint32_t *)(&(THM_BASE.instance[i]));
adev->reg_offset[CLK_HWIP][i] = (uint32_t *)(&(CLK_BASE.instance[i]));
}
return 0;
}

4806
drivers/gpu/drm/amd/amdgpu/navi10_sdma_pkt_open.h Normal file
Просмотреть файл

Разница между файлами не показана из-за своего большого размера Загрузить разницу

334
drivers/gpu/drm/amd/amdgpu/nbio_v2_3.c Normal file
Просмотреть файл

@ -0,0 +1,334 @@
/*
* Copyright 2019 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.
*
*/
#include "amdgpu.h"
#include "amdgpu_atombios.h"
#include "nbio_v2_3.h"
#include "nbio/nbio_2_3_default.h"
#include "nbio/nbio_2_3_offset.h"
#include "nbio/nbio_2_3_sh_mask.h"
#define smnPCIE_CONFIG_CNTL 0x11180044
#define smnCPM_CONTROL 0x11180460
#define smnPCIE_CNTL2 0x11180070
static u32 nbio_v2_3_get_rev_id(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
tmp &= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0_MASK;
tmp >>= RCC_DEV0_EPF0_STRAP0__STRAP_ATI_REV_ID_DEV0_F0__SHIFT;
return tmp;
}
static void nbio_v2_3_mc_access_enable(struct amdgpu_device *adev, bool enable)
{
if (enable)
WREG32_SOC15(NBIO, 0, mmBIF_FB_EN,
BIF_FB_EN__FB_READ_EN_MASK |
BIF_FB_EN__FB_WRITE_EN_MASK);
else
WREG32_SOC15(NBIO, 0, mmBIF_FB_EN, 0);
}
static void nbio_v2_3_hdp_flush(struct amdgpu_device *adev,
struct amdgpu_ring *ring)
{
if (!ring || !ring->funcs->emit_wreg)
WREG32_SOC15_NO_KIQ(NBIO, 0, mmBIF_BX_PF_HDP_MEM_COHERENCY_FLUSH_CNTL, 0);
else
amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
NBIO, 0, mmBIF_BX_PF_HDP_MEM_COHERENCY_FLUSH_CNTL), 0);
}
static u32 nbio_v2_3_get_memsize(struct amdgpu_device *adev)
{
return RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_RCC_CONFIG_MEMSIZE);
}
static void nbio_v2_3_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
bool use_doorbell, int doorbell_index,
int doorbell_size)
{
u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
u32 doorbell_range = RREG32(reg);
if (use_doorbell) {
doorbell_range = REG_SET_FIELD(doorbell_range,
BIF_SDMA0_DOORBELL_RANGE, OFFSET,
doorbell_index);
doorbell_range = REG_SET_FIELD(doorbell_range,
BIF_SDMA0_DOORBELL_RANGE, SIZE,
doorbell_size);
} else
doorbell_range = REG_SET_FIELD(doorbell_range,
BIF_SDMA0_DOORBELL_RANGE, SIZE,
0);
WREG32(reg, doorbell_range);
}
static void nbio_v2_3_vcn_doorbell_range(struct amdgpu_device *adev, bool use_doorbell,
int doorbell_index)
{
u32 reg = SOC15_REG_OFFSET(NBIO, 0, mmBIF_MMSCH0_DOORBELL_RANGE);
u32 doorbell_range = RREG32(reg);
if (use_doorbell) {
doorbell_range = REG_SET_FIELD(doorbell_range,
BIF_MMSCH0_DOORBELL_RANGE, OFFSET,
doorbell_index);
doorbell_range = REG_SET_FIELD(doorbell_range,
BIF_MMSCH0_DOORBELL_RANGE, SIZE, 8);
} else
doorbell_range = REG_SET_FIELD(doorbell_range,
BIF_MMSCH0_DOORBELL_RANGE, SIZE, 0);
WREG32(reg, doorbell_range);
}
static void nbio_v2_3_enable_doorbell_aperture(struct amdgpu_device *adev,
bool enable)
{
WREG32_FIELD15(NBIO, 0, RCC_DEV0_EPF0_RCC_DOORBELL_APER_EN, BIF_DOORBELL_APER_EN,
enable ? 1 : 0);
}
static void nbio_v2_3_enable_doorbell_selfring_aperture(struct amdgpu_device *adev,
bool enable)
{
u32 tmp = 0;
if (enable) {
tmp = REG_SET_FIELD(tmp, BIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
DOORBELL_SELFRING_GPA_APER_EN, 1) |
REG_SET_FIELD(tmp, BIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
DOORBELL_SELFRING_GPA_APER_MODE, 1) |
REG_SET_FIELD(tmp, BIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
DOORBELL_SELFRING_GPA_APER_SIZE, 0);
WREG32_SOC15(NBIO, 0, mmBIF_BX_PF_DOORBELL_SELFRING_GPA_APER_BASE_LOW,
lower_32_bits(adev->doorbell.base));
WREG32_SOC15(NBIO, 0, mmBIF_BX_PF_DOORBELL_SELFRING_GPA_APER_BASE_HIGH,
upper_32_bits(adev->doorbell.base));
}
WREG32_SOC15(NBIO, 0, mmBIF_BX_PF_DOORBELL_SELFRING_GPA_APER_CNTL,
tmp);
}
static void nbio_v2_3_ih_doorbell_range(struct amdgpu_device *adev,
bool use_doorbell, int doorbell_index)
{
u32 ih_doorbell_range = RREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE);
if (use_doorbell) {
ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
BIF_IH_DOORBELL_RANGE, OFFSET,
doorbell_index);
ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
BIF_IH_DOORBELL_RANGE, SIZE,
2);
} else
ih_doorbell_range = REG_SET_FIELD(ih_doorbell_range,
BIF_IH_DOORBELL_RANGE, SIZE,
0);
WREG32_SOC15(NBIO, 0, mmBIF_IH_DOORBELL_RANGE, ih_doorbell_range);
}
static void nbio_v2_3_ih_control(struct amdgpu_device *adev)
{
u32 interrupt_cntl;
/* setup interrupt control */
WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL2, adev->dummy_page_addr >> 8);
interrupt_cntl = RREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL);
/*
* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=0 - dummy read disabled with msi, enabled without msi
* INTERRUPT_CNTL__IH_DUMMY_RD_OVERRIDE_MASK=1 - dummy read controlled by IH_DUMMY_RD_EN
*/
interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL,
IH_DUMMY_RD_OVERRIDE, 0);
/* INTERRUPT_CNTL__IH_REQ_NONSNOOP_EN_MASK=1 if ring is in non-cacheable memory, e.g., vram */
interrupt_cntl = REG_SET_FIELD(interrupt_cntl, INTERRUPT_CNTL,
IH_REQ_NONSNOOP_EN, 0);
WREG32_SOC15(NBIO, 0, mmINTERRUPT_CNTL, interrupt_cntl);
}
static void nbio_v2_3_update_medium_grain_clock_gating(struct amdgpu_device *adev,
bool enable)
{
uint32_t def, data;
def = data = RREG32_PCIE(smnCPM_CONTROL);
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_MGCG)) {
data |= (CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
} else {
data &= ~(CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_DYN_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_LCNT_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_REGS_GATE_ENABLE_MASK |
CPM_CONTROL__TXCLK_PRBS_GATE_ENABLE_MASK |
CPM_CONTROL__REFCLK_REGS_GATE_ENABLE_MASK);
}
if (def != data)
WREG32_PCIE(smnCPM_CONTROL, data);
}
static void nbio_v2_3_update_medium_grain_light_sleep(struct amdgpu_device *adev,
bool enable)
{
uint32_t def, data;
def = data = RREG32_PCIE(smnPCIE_CNTL2);
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_BIF_LS)) {
data |= (PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
PCIE_CNTL2__MST_MEM_LS_EN_MASK |
PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
} else {
data &= ~(PCIE_CNTL2__SLV_MEM_LS_EN_MASK |
PCIE_CNTL2__MST_MEM_LS_EN_MASK |
PCIE_CNTL2__REPLAY_MEM_LS_EN_MASK);
}
if (def != data)
WREG32_PCIE(smnPCIE_CNTL2, data);
}
static void nbio_v2_3_get_clockgating_state(struct amdgpu_device *adev,
u32 *flags)
{
int data;
/* AMD_CG_SUPPORT_BIF_MGCG */
data = RREG32_PCIE(smnCPM_CONTROL);
if (data & CPM_CONTROL__LCLK_DYN_GATE_ENABLE_MASK)
*flags |= AMD_CG_SUPPORT_BIF_MGCG;
/* AMD_CG_SUPPORT_BIF_LS */
data = RREG32_PCIE(smnPCIE_CNTL2);
if (data & PCIE_CNTL2__SLV_MEM_LS_EN_MASK)
*flags |= AMD_CG_SUPPORT_BIF_LS;
}
static u32 nbio_v2_3_get_hdp_flush_req_offset(struct amdgpu_device *adev)
{
return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF_GPU_HDP_FLUSH_REQ);
}
static u32 nbio_v2_3_get_hdp_flush_done_offset(struct amdgpu_device *adev)
{
return SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF_GPU_HDP_FLUSH_DONE);
}
static u32 nbio_v2_3_get_pcie_index_offset(struct amdgpu_device *adev)
{
return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_INDEX2);
}
static u32 nbio_v2_3_get_pcie_data_offset(struct amdgpu_device *adev)
{
return SOC15_REG_OFFSET(NBIO, 0, mmPCIE_DATA2);
}
const struct nbio_hdp_flush_reg nbio_v2_3_hdp_flush_reg = {
.ref_and_mask_cp0 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP0_MASK,
.ref_and_mask_cp1 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP1_MASK,
.ref_and_mask_cp2 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP2_MASK,
.ref_and_mask_cp3 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP3_MASK,
.ref_and_mask_cp4 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP4_MASK,
.ref_and_mask_cp5 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP5_MASK,
.ref_and_mask_cp6 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP6_MASK,
.ref_and_mask_cp7 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP7_MASK,
.ref_and_mask_cp8 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP8_MASK,
.ref_and_mask_cp9 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__CP9_MASK,
.ref_and_mask_sdma0 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__SDMA0_MASK,
.ref_and_mask_sdma1 = BIF_BX_PF_GPU_HDP_FLUSH_DONE__SDMA1_MASK,
};
static void nbio_v2_3_detect_hw_virt(struct amdgpu_device *adev)
{
uint32_t reg;
reg = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_RCC_IOV_FUNC_IDENTIFIER);
if (reg & 1)
adev->virt.caps |= AMDGPU_SRIOV_CAPS_IS_VF;
if (reg & 0x80000000)
adev->virt.caps |= AMDGPU_SRIOV_CAPS_ENABLE_IOV;
if (!reg) {
if (is_virtual_machine()) /* passthrough mode exclus sriov mod */
adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
}
}
static void nbio_v2_3_init_registers(struct amdgpu_device *adev)
{
uint32_t def, data;
def = data = RREG32_PCIE(smnPCIE_CONFIG_CNTL);
data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_MODE, 1);
data = REG_SET_FIELD(data, PCIE_CONFIG_CNTL, CI_SWUS_MAX_READ_REQUEST_SIZE_PRIV, 1);
if (def != data)
WREG32_PCIE(smnPCIE_CONFIG_CNTL, data);
}
const struct amdgpu_nbio_funcs nbio_v2_3_funcs = {
.hdp_flush_reg = &nbio_v2_3_hdp_flush_reg,
.get_hdp_flush_req_offset = nbio_v2_3_get_hdp_flush_req_offset,
.get_hdp_flush_done_offset = nbio_v2_3_get_hdp_flush_done_offset,
.get_pcie_index_offset = nbio_v2_3_get_pcie_index_offset,
.get_pcie_data_offset = nbio_v2_3_get_pcie_data_offset,
.get_rev_id = nbio_v2_3_get_rev_id,
.mc_access_enable = nbio_v2_3_mc_access_enable,
.hdp_flush = nbio_v2_3_hdp_flush,
.get_memsize = nbio_v2_3_get_memsize,
.sdma_doorbell_range = nbio_v2_3_sdma_doorbell_range,
.vcn_doorbell_range = nbio_v2_3_vcn_doorbell_range,
.enable_doorbell_aperture = nbio_v2_3_enable_doorbell_aperture,
.enable_doorbell_selfring_aperture = nbio_v2_3_enable_doorbell_selfring_aperture,
.ih_doorbell_range = nbio_v2_3_ih_doorbell_range,
.update_medium_grain_clock_gating = nbio_v2_3_update_medium_grain_clock_gating,
.update_medium_grain_light_sleep = nbio_v2_3_update_medium_grain_light_sleep,
.get_clockgating_state = nbio_v2_3_get_clockgating_state,
.ih_control = nbio_v2_3_ih_control,
.init_registers = nbio_v2_3_init_registers,
.detect_hw_virt = nbio_v2_3_detect_hw_virt,
};

31
drivers/gpu/drm/amd/amdgpu/nbio_v2_3.h Normal file
Просмотреть файл

@ -0,0 +1,31 @@
/*
* Copyright 2019 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 __NBIO_V2_3_H__
#define __NBIO_V2_3_H__
#include "soc15_common.h"
extern const struct amdgpu_nbio_funcs nbio_v2_3_funcs;
#endif

778
drivers/gpu/drm/amd/amdgpu/nv.c Normal file
Просмотреть файл

@ -0,0 +1,778 @@
/*
* Copyright 2019 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.
*
*/
#include <linux/firmware.h>
#include <linux/slab.h>
#include <linux/module.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_atombios.h"
#include "amdgpu_ih.h"
#include "amdgpu_uvd.h"
#include "amdgpu_vce.h"
#include "amdgpu_ucode.h"
#include "amdgpu_psp.h"
#include "atom.h"
#include "amd_pcie.h"
#include "gc/gc_10_1_0_offset.h"
#include "gc/gc_10_1_0_sh_mask.h"
#include "hdp/hdp_5_0_0_offset.h"
#include "hdp/hdp_5_0_0_sh_mask.h"
#include "soc15.h"
#include "soc15_common.h"
#include "gmc_v10_0.h"
#include "gfxhub_v2_0.h"
#include "mmhub_v2_0.h"
#include "nv.h"
#include "navi10_ih.h"
#include "gfx_v10_0.h"
#include "sdma_v5_0.h"
#include "vcn_v2_0.h"
#include "dce_virtual.h"
#include "mes_v10_1.h"
static const struct amd_ip_funcs nv_common_ip_funcs;
/*
* Indirect registers accessor
*/
static u32 nv_pcie_rreg(struct amdgpu_device *adev, u32 reg)
{
unsigned long flags, address, data;
u32 r;
address = adev->nbio_funcs->get_pcie_index_offset(adev);
data = adev->nbio_funcs->get_pcie_data_offset(adev);
spin_lock_irqsave(&adev->pcie_idx_lock, flags);
WREG32(address, reg);
(void)RREG32(address);
r = RREG32(data);
spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
return r;
}
static void nv_pcie_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
{
unsigned long flags, address, data;
address = adev->nbio_funcs->get_pcie_index_offset(adev);
data = adev->nbio_funcs->get_pcie_data_offset(adev);
spin_lock_irqsave(&adev->pcie_idx_lock, flags);
WREG32(address, reg);
(void)RREG32(address);
WREG32(data, v);
(void)RREG32(data);
spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
}
static u32 nv_didt_rreg(struct amdgpu_device *adev, u32 reg)
{
unsigned long flags, address, data;
u32 r;
address = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_INDEX);
data = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_DATA);
spin_lock_irqsave(&adev->didt_idx_lock, flags);
WREG32(address, (reg));
r = RREG32(data);
spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
return r;
}
static void nv_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
{
unsigned long flags, address, data;
address = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_INDEX);
data = SOC15_REG_OFFSET(GC, 0, mmDIDT_IND_DATA);
spin_lock_irqsave(&adev->didt_idx_lock, flags);
WREG32(address, (reg));
WREG32(data, (v));
spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
}
static u32 nv_get_config_memsize(struct amdgpu_device *adev)
{
return adev->nbio_funcs->get_memsize(adev);
}
static u32 nv_get_xclk(struct amdgpu_device *adev)
{
return adev->clock.spll.reference_freq;
}
void nv_grbm_select(struct amdgpu_device *adev,
u32 me, u32 pipe, u32 queue, u32 vmid)
{
u32 grbm_gfx_cntl = 0;
grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, PIPEID, pipe);
grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, MEID, me);
grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, VMID, vmid);
grbm_gfx_cntl = REG_SET_FIELD(grbm_gfx_cntl, GRBM_GFX_CNTL, QUEUEID, queue);
WREG32(SOC15_REG_OFFSET(GC, 0, mmGRBM_GFX_CNTL), grbm_gfx_cntl);
}
static void nv_vga_set_state(struct amdgpu_device *adev, bool state)
{
/* todo */
}
static bool nv_read_disabled_bios(struct amdgpu_device *adev)
{
/* todo */
return false;
}
static bool nv_read_bios_from_rom(struct amdgpu_device *adev,
u8 *bios, u32 length_bytes)
{
/* TODO: will implement it when SMU header is available */
return false;
}
static struct soc15_allowed_register_entry nv_allowed_read_registers[] = {
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)},
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)},
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE0)},
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE1)},
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE2)},
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS_SE3)},
#if 0 /* TODO: will set it when SDMA header is available */
{ SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_STATUS_REG)},
{ SOC15_REG_ENTRY(SDMA1, 0, mmSDMA1_STATUS_REG)},
#endif
{ SOC15_REG_ENTRY(GC, 0, mmCP_STAT)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT1)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT2)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_STALLED_STAT3)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_BUSY_STAT)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STALLED_STAT1)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_CPF_STATUS)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STALLED_STAT1)},
{ SOC15_REG_ENTRY(GC, 0, mmCP_CPC_STATUS)},
{ SOC15_REG_ENTRY(GC, 0, mmGB_ADDR_CONFIG)},
};
static uint32_t nv_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
u32 sh_num, u32 reg_offset)
{
uint32_t val;
mutex_lock(&adev->grbm_idx_mutex);
if (se_num != 0xffffffff || sh_num != 0xffffffff)
amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
val = RREG32(reg_offset);
if (se_num != 0xffffffff || sh_num != 0xffffffff)
amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
mutex_unlock(&adev->grbm_idx_mutex);
return val;
}
static uint32_t nv_get_register_value(struct amdgpu_device *adev,
bool indexed, u32 se_num,
u32 sh_num, u32 reg_offset)
{
if (indexed) {
return nv_read_indexed_register(adev, se_num, sh_num, reg_offset);
} else {
if (reg_offset == SOC15_REG_OFFSET(GC, 0, mmGB_ADDR_CONFIG))
return adev->gfx.config.gb_addr_config;
return RREG32(reg_offset);
}
}
static int nv_read_register(struct amdgpu_device *adev, u32 se_num,
u32 sh_num, u32 reg_offset, u32 *value)
{
uint32_t i;
struct soc15_allowed_register_entry *en;
*value = 0;
for (i = 0; i < ARRAY_SIZE(nv_allowed_read_registers); i++) {
en = &nv_allowed_read_registers[i];
if (reg_offset !=
(adev->reg_offset[en->hwip][en->inst][en->seg] + en->reg_offset))
continue;
*value = nv_get_register_value(adev,
nv_allowed_read_registers[i].grbm_indexed,
se_num, sh_num, reg_offset);
return 0;
}
return -EINVAL;
}
#if 0
static void nv_gpu_pci_config_reset(struct amdgpu_device *adev)
{
u32 i;
dev_info(adev->dev, "GPU pci config reset\n");
/* disable BM */
pci_clear_master(adev->pdev);
/* reset */
amdgpu_pci_config_reset(adev);
udelay(100);
/* wait for asic to come out of reset */
for (i = 0; i < adev->usec_timeout; i++) {
u32 memsize = nbio_v2_3_get_memsize(adev);
if (memsize != 0xffffffff)
break;
udelay(1);
}
}
#endif
static int nv_asic_reset(struct amdgpu_device *adev)
{
/* FIXME: it doesn't work since vega10 */
#if 0
amdgpu_atombios_scratch_regs_engine_hung(adev, true);
nv_gpu_pci_config_reset(adev);
amdgpu_atombios_scratch_regs_engine_hung(adev, false);
#endif
return 0;
}
static int nv_set_uvd_clocks(struct amdgpu_device *adev, u32 vclk, u32 dclk)
{
/* todo */
return 0;
}
static int nv_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk)
{
/* todo */
return 0;
}
static void nv_pcie_gen3_enable(struct amdgpu_device *adev)
{
if (pci_is_root_bus(adev->pdev->bus))
return;
if (amdgpu_pcie_gen2 == 0)
return;
if (!(adev->pm.pcie_gen_mask & (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)))
return;
/* todo */
}
static void nv_program_aspm(struct amdgpu_device *adev)
{
if (amdgpu_aspm == 0)
return;
/* todo */
}
static void nv_enable_doorbell_aperture(struct amdgpu_device *adev,
bool enable)
{
adev->nbio_funcs->enable_doorbell_aperture(adev, enable);
adev->nbio_funcs->enable_doorbell_selfring_aperture(adev, enable);
}
static const struct amdgpu_ip_block_version nv_common_ip_block =
{
.type = AMD_IP_BLOCK_TYPE_COMMON,
.major = 1,
.minor = 0,
.rev = 0,
.funcs = &nv_common_ip_funcs,
};
int nv_set_ip_blocks(struct amdgpu_device *adev)
{
/* Set IP register base before any HW register access */
switch (adev->asic_type) {
case CHIP_NAVI10:
navi10_reg_base_init(adev);
break;
default:
return -EINVAL;
}
adev->nbio_funcs = &nbio_v2_3_funcs;
adev->nbio_funcs->detect_hw_virt(adev);
switch (adev->asic_type) {
case CHIP_NAVI10:
amdgpu_device_ip_block_add(adev, &nv_common_ip_block);
amdgpu_device_ip_block_add(adev, &gmc_v10_0_ip_block);
amdgpu_device_ip_block_add(adev, &navi10_ih_ip_block);
amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block);
if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP &&
is_support_sw_smu(adev))
amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
if (adev->enable_virtual_display || amdgpu_sriov_vf(adev))
amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
else if (amdgpu_device_has_dc_support(adev))
amdgpu_device_ip_block_add(adev, &dm_ip_block);
amdgpu_device_ip_block_add(adev, &gfx_v10_0_ip_block);
amdgpu_device_ip_block_add(adev, &sdma_v5_0_ip_block);
if (adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT &&
is_support_sw_smu(adev))
amdgpu_device_ip_block_add(adev, &smu_v11_0_ip_block);
amdgpu_device_ip_block_add(adev, &vcn_v2_0_ip_block);
if (adev->enable_mes)
amdgpu_device_ip_block_add(adev, &mes_v10_1_ip_block);
break;
default:
return -EINVAL;
}
return 0;
}
static uint32_t nv_get_rev_id(struct amdgpu_device *adev)
{
return adev->nbio_funcs->get_rev_id(adev);
}
static void nv_flush_hdp(struct amdgpu_device *adev, struct amdgpu_ring *ring)
{
adev->nbio_funcs->hdp_flush(adev, ring);
}
static void nv_invalidate_hdp(struct amdgpu_device *adev,
struct amdgpu_ring *ring)
{
if (!ring || !ring->funcs->emit_wreg) {
WREG32_SOC15_NO_KIQ(NBIO, 0, mmHDP_READ_CACHE_INVALIDATE, 1);
} else {
amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
}
}
static bool nv_need_full_reset(struct amdgpu_device *adev)
{
return true;
}
static void nv_get_pcie_usage(struct amdgpu_device *adev,
uint64_t *count0,
uint64_t *count1)
{
/*TODO*/
}
static bool nv_need_reset_on_init(struct amdgpu_device *adev)
{
#if 0
u32 sol_reg;
if (adev->flags & AMD_IS_APU)
return false;
/* Check sOS sign of life register to confirm sys driver and sOS
* are already been loaded.
*/
sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
if (sol_reg)
return true;
#endif
/* TODO: re-enable it when mode1 reset is functional */
return false;
}
static void nv_init_doorbell_index(struct amdgpu_device *adev)
{
adev->doorbell_index.kiq = AMDGPU_NAVI10_DOORBELL_KIQ;
adev->doorbell_index.mec_ring0 = AMDGPU_NAVI10_DOORBELL_MEC_RING0;
adev->doorbell_index.mec_ring1 = AMDGPU_NAVI10_DOORBELL_MEC_RING1;
adev->doorbell_index.mec_ring2 = AMDGPU_NAVI10_DOORBELL_MEC_RING2;
adev->doorbell_index.mec_ring3 = AMDGPU_NAVI10_DOORBELL_MEC_RING3;
adev->doorbell_index.mec_ring4 = AMDGPU_NAVI10_DOORBELL_MEC_RING4;
adev->doorbell_index.mec_ring5 = AMDGPU_NAVI10_DOORBELL_MEC_RING5;
adev->doorbell_index.mec_ring6 = AMDGPU_NAVI10_DOORBELL_MEC_RING6;
adev->doorbell_index.mec_ring7 = AMDGPU_NAVI10_DOORBELL_MEC_RING7;
adev->doorbell_index.userqueue_start = AMDGPU_NAVI10_DOORBELL_USERQUEUE_START;
adev->doorbell_index.userqueue_end = AMDGPU_NAVI10_DOORBELL_USERQUEUE_END;
adev->doorbell_index.gfx_ring0 = AMDGPU_NAVI10_DOORBELL_GFX_RING0;
adev->doorbell_index.gfx_ring1 = AMDGPU_NAVI10_DOORBELL_GFX_RING1;
adev->doorbell_index.sdma_engine[0] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine[1] = AMDGPU_NAVI10_DOORBELL_sDMA_ENGINE1;
adev->doorbell_index.ih = AMDGPU_NAVI10_DOORBELL_IH;
adev->doorbell_index.vcn.vcn_ring0_1 = AMDGPU_NAVI10_DOORBELL64_VCN0_1;
adev->doorbell_index.vcn.vcn_ring2_3 = AMDGPU_NAVI10_DOORBELL64_VCN2_3;
adev->doorbell_index.vcn.vcn_ring4_5 = AMDGPU_NAVI10_DOORBELL64_VCN4_5;
adev->doorbell_index.vcn.vcn_ring6_7 = AMDGPU_NAVI10_DOORBELL64_VCN6_7;
adev->doorbell_index.first_non_cp = AMDGPU_NAVI10_DOORBELL64_FIRST_NON_CP;
adev->doorbell_index.last_non_cp = AMDGPU_NAVI10_DOORBELL64_LAST_NON_CP;
adev->doorbell_index.max_assignment = AMDGPU_NAVI10_DOORBELL_MAX_ASSIGNMENT << 1;
adev->doorbell_index.sdma_doorbell_range = 20;
}
static const struct amdgpu_asic_funcs nv_asic_funcs =
{
.read_disabled_bios = &nv_read_disabled_bios,
.read_bios_from_rom = &nv_read_bios_from_rom,
.read_register = &nv_read_register,
.reset = &nv_asic_reset,
.set_vga_state = &nv_vga_set_state,
.get_xclk = &nv_get_xclk,
.set_uvd_clocks = &nv_set_uvd_clocks,
.set_vce_clocks = &nv_set_vce_clocks,
.get_config_memsize = &nv_get_config_memsize,
.flush_hdp = &nv_flush_hdp,
.invalidate_hdp = &nv_invalidate_hdp,
.init_doorbell_index = &nv_init_doorbell_index,
.need_full_reset = &nv_need_full_reset,
.get_pcie_usage = &nv_get_pcie_usage,
.need_reset_on_init = &nv_need_reset_on_init,
};
static int nv_common_early_init(void *handle)
{
bool psp_enabled = false;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
adev->smc_rreg = NULL;
adev->smc_wreg = NULL;
adev->pcie_rreg = &nv_pcie_rreg;
adev->pcie_wreg = &nv_pcie_wreg;
/* TODO: will add them during VCN v2 implementation */
adev->uvd_ctx_rreg = NULL;
adev->uvd_ctx_wreg = NULL;
adev->didt_rreg = &nv_didt_rreg;
adev->didt_wreg = &nv_didt_wreg;
adev->asic_funcs = &nv_asic_funcs;
if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_PSP) &&
(amdgpu_ip_block_mask & (1 << AMD_IP_BLOCK_TYPE_PSP)))
psp_enabled = true;
adev->rev_id = nv_get_rev_id(adev);
adev->external_rev_id = 0xff;
switch (adev->asic_type) {
case CHIP_NAVI10:
adev->cg_flags = AMD_CG_SUPPORT_GFX_MGCG |
AMD_CG_SUPPORT_GFX_CGCG |
AMD_CG_SUPPORT_IH_CG |
AMD_CG_SUPPORT_HDP_MGCG |
AMD_CG_SUPPORT_HDP_LS |
AMD_CG_SUPPORT_SDMA_MGCG |
AMD_CG_SUPPORT_SDMA_LS |
AMD_CG_SUPPORT_MC_MGCG |
AMD_CG_SUPPORT_MC_LS |
AMD_CG_SUPPORT_ATHUB_MGCG |
AMD_CG_SUPPORT_ATHUB_LS |
AMD_CG_SUPPORT_VCN_MGCG |
AMD_CG_SUPPORT_BIF_MGCG |
AMD_CG_SUPPORT_BIF_LS;
adev->pg_flags = AMD_PG_SUPPORT_VCN |
AMD_PG_SUPPORT_VCN_DPG |
AMD_PG_SUPPORT_MMHUB |
AMD_PG_SUPPORT_ATHUB;
adev->external_rev_id = adev->rev_id + 0x1;
break;
default:
/* FIXME: not supported yet */
return -EINVAL;
}
return 0;
}
static int nv_common_late_init(void *handle)
{
return 0;
}
static int nv_common_sw_init(void *handle)
{
return 0;
}
static int nv_common_sw_fini(void *handle)
{
return 0;
}
static int nv_common_hw_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
/* enable pcie gen2/3 link */
nv_pcie_gen3_enable(adev);
/* enable aspm */
nv_program_aspm(adev);
/* setup nbio registers */
adev->nbio_funcs->init_registers(adev);
/* enable the doorbell aperture */
nv_enable_doorbell_aperture(adev, true);
return 0;
}
static int nv_common_hw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
/* disable the doorbell aperture */
nv_enable_doorbell_aperture(adev, false);
return 0;
}
static int nv_common_suspend(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
return nv_common_hw_fini(adev);
}
static int nv_common_resume(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
return nv_common_hw_init(adev);
}
static bool nv_common_is_idle(void *handle)
{
return true;
}
static int nv_common_wait_for_idle(void *handle)
{
return 0;
}
static int nv_common_soft_reset(void *handle)
{
return 0;
}
static void nv_update_hdp_mem_power_gating(struct amdgpu_device *adev,
bool enable)
{
uint32_t hdp_clk_cntl, hdp_clk_cntl1;
uint32_t hdp_mem_pwr_cntl;
if (!(adev->cg_flags & (AMD_CG_SUPPORT_HDP_LS |
AMD_CG_SUPPORT_HDP_DS |
AMD_CG_SUPPORT_HDP_SD)))
return;
hdp_clk_cntl = hdp_clk_cntl1 = RREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL);
hdp_mem_pwr_cntl = RREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL);
/* Before doing clock/power mode switch,
* forced on IPH & RC clock */
hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
IPH_MEM_CLK_SOFT_OVERRIDE, 1);
hdp_clk_cntl = REG_SET_FIELD(hdp_clk_cntl, HDP_CLK_CNTL,
RC_MEM_CLK_SOFT_OVERRIDE, 1);
WREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL, hdp_clk_cntl);
/* HDP 5.0 doesn't support dynamic power mode switch,
* disable clock and power gating before any changing */
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_CTRL_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_LS_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_DS_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_SD_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
RC_MEM_POWER_CTRL_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
RC_MEM_POWER_LS_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
RC_MEM_POWER_DS_EN, 0);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl, HDP_MEM_POWER_CTRL,
RC_MEM_POWER_SD_EN, 0);
WREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
/* only one clock gating mode (LS/DS/SD) can be enabled */
if (adev->cg_flags & AMD_CG_SUPPORT_HDP_LS) {
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_LS_EN, enable);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
HDP_MEM_POWER_CTRL,
RC_MEM_POWER_LS_EN, enable);
} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_DS) {
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_DS_EN, enable);
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
HDP_MEM_POWER_CTRL,
RC_MEM_POWER_DS_EN, enable);
} else if (adev->cg_flags & AMD_CG_SUPPORT_HDP_SD) {
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
HDP_MEM_POWER_CTRL,
IPH_MEM_POWER_SD_EN, enable);
/* RC should not use shut down mode, fallback to ds */
hdp_mem_pwr_cntl = REG_SET_FIELD(hdp_mem_pwr_cntl,
HDP_MEM_POWER_CTRL,
RC_MEM_POWER_DS_EN, enable);
}
WREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL, hdp_mem_pwr_cntl);
/* restore IPH & RC clock override after clock/power mode changing */
WREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL, hdp_clk_cntl1);
}
static void nv_update_hdp_clock_gating(struct amdgpu_device *adev,
bool enable)
{
uint32_t hdp_clk_cntl;
if (!(adev->cg_flags & AMD_CG_SUPPORT_HDP_MGCG))
return;
hdp_clk_cntl = RREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL);
if (enable) {
hdp_clk_cntl &=
~(uint32_t)
(HDP_CLK_CNTL__IPH_MEM_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK);
} else {
hdp_clk_cntl |= HDP_CLK_CNTL__IPH_MEM_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK;
}
WREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL, hdp_clk_cntl);
}
static int nv_common_set_clockgating_state(void *handle,
enum amd_clockgating_state state)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (amdgpu_sriov_vf(adev))
return 0;
switch (adev->asic_type) {
case CHIP_NAVI10:
adev->nbio_funcs->update_medium_grain_clock_gating(adev,
state == AMD_CG_STATE_GATE ? true : false);
adev->nbio_funcs->update_medium_grain_light_sleep(adev,
state == AMD_CG_STATE_GATE ? true : false);
nv_update_hdp_mem_power_gating(adev,
state == AMD_CG_STATE_GATE ? true : false);
nv_update_hdp_clock_gating(adev,
state == AMD_CG_STATE_GATE ? true : false);
break;
default:
break;
}
return 0;
}
static int nv_common_set_powergating_state(void *handle,
enum amd_powergating_state state)
{
/* TODO */
return 0;
}
static void nv_common_get_clockgating_state(void *handle, u32 *flags)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
uint32_t tmp;
if (amdgpu_sriov_vf(adev))
*flags = 0;
adev->nbio_funcs->get_clockgating_state(adev, flags);
/* AMD_CG_SUPPORT_HDP_MGCG */
tmp = RREG32_SOC15(HDP, 0, mmHDP_CLK_CNTL);
if (!(tmp & (HDP_CLK_CNTL__IPH_MEM_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__RC_MEM_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__DBUS_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__DYN_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__XDP_REG_CLK_SOFT_OVERRIDE_MASK |
HDP_CLK_CNTL__HDP_REG_CLK_SOFT_OVERRIDE_MASK)))
*flags |= AMD_CG_SUPPORT_HDP_MGCG;
/* AMD_CG_SUPPORT_HDP_LS/DS/SD */
tmp = RREG32_SOC15(HDP, 0, mmHDP_MEM_POWER_CTRL);
if (tmp & HDP_MEM_POWER_CTRL__IPH_MEM_POWER_LS_EN_MASK)
*flags |= AMD_CG_SUPPORT_HDP_LS;
else if (tmp & HDP_MEM_POWER_CTRL__IPH_MEM_POWER_DS_EN_MASK)
*flags |= AMD_CG_SUPPORT_HDP_DS;
else if (tmp & HDP_MEM_POWER_CTRL__IPH_MEM_POWER_SD_EN_MASK)
*flags |= AMD_CG_SUPPORT_HDP_SD;
return;
}
static const struct amd_ip_funcs nv_common_ip_funcs = {
.name = "nv_common",
.early_init = nv_common_early_init,
.late_init = nv_common_late_init,
.sw_init = nv_common_sw_init,
.sw_fini = nv_common_sw_fini,
.hw_init = nv_common_hw_init,
.hw_fini = nv_common_hw_fini,
.suspend = nv_common_suspend,
.resume = nv_common_resume,
.is_idle = nv_common_is_idle,
.wait_for_idle = nv_common_wait_for_idle,
.soft_reset = nv_common_soft_reset,
.set_clockgating_state = nv_common_set_clockgating_state,
.set_powergating_state = nv_common_set_powergating_state,
.get_clockgating_state = nv_common_get_clockgating_state,
};

33
drivers/gpu/drm/amd/amdgpu/nv.h Normal file
Просмотреть файл

@ -0,0 +1,33 @@
/*
* Copyright 2019 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 __NV_H__
#define __NV_H__
#include "nbio_v2_3.h"
void nv_grbm_select(struct amdgpu_device *adev,
u32 me, u32 pipe, u32 queue, u32 vmid);
int nv_set_ip_blocks(struct amdgpu_device *adev);
int navi10_reg_base_init(struct amdgpu_device *adev);
#endif

418
drivers/gpu/drm/amd/amdgpu/nvd.h Normal file
Просмотреть файл

@ -0,0 +1,418 @@
/*
* Copyright 2019 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 NVD_H
#define NVD_H
/**
* Navi's PM4 definitions
*/
#define PACKET_TYPE0 0
#define PACKET_TYPE1 1
#define PACKET_TYPE2 2
#define PACKET_TYPE3 3
#define CP_PACKET_GET_TYPE(h) (((h) >> 30) & 3)
#define CP_PACKET_GET_COUNT(h) (((h) >> 16) & 0x3FFF)
#define CP_PACKET0_GET_REG(h) ((h) & 0xFFFF)
#define CP_PACKET3_GET_OPCODE(h) (((h) >> 8) & 0xFF)
#define PACKET0(reg, n) ((PACKET_TYPE0 << 30) | \
((reg) & 0xFFFF) | \
((n) & 0x3FFF) << 16)
#define CP_PACKET2 0x80000000
#define PACKET2_PAD_SHIFT 0
#define PACKET2_PAD_MASK (0x3fffffff << 0)
#define PACKET2(v) (CP_PACKET2 | REG_SET(PACKET2_PAD, (v)))
#define PACKET3(op, n) ((PACKET_TYPE3 << 30) | \
(((op) & 0xFF) << 8) | \
((n) & 0x3FFF) << 16)
#define PACKET3_COMPUTE(op, n) (PACKET3(op, n) | 1 << 1)
/* Packet 3 types */
#define PACKET3_NOP 0x10
#define PACKET3_SET_BASE 0x11
#define PACKET3_BASE_INDEX(x) ((x) << 0)
#define CE_PARTITION_BASE 3
#define PACKET3_CLEAR_STATE 0x12
#define PACKET3_INDEX_BUFFER_SIZE 0x13
#define PACKET3_DISPATCH_DIRECT 0x15
#define PACKET3_DISPATCH_INDIRECT 0x16
#define PACKET3_INDIRECT_BUFFER_END 0x17
#define PACKET3_INDIRECT_BUFFER_CNST_END 0x19
#define PACKET3_ATOMIC_GDS 0x1D
#define PACKET3_ATOMIC_MEM 0x1E
#define PACKET3_OCCLUSION_QUERY 0x1F
#define PACKET3_SET_PREDICATION 0x20
#define PACKET3_REG_RMW 0x21
#define PACKET3_COND_EXEC 0x22
#define PACKET3_PRED_EXEC 0x23
#define PACKET3_DRAW_INDIRECT 0x24
#define PACKET3_DRAW_INDEX_INDIRECT 0x25
#define PACKET3_INDEX_BASE 0x26
#define PACKET3_DRAW_INDEX_2 0x27
#define PACKET3_CONTEXT_CONTROL 0x28
#define PACKET3_INDEX_TYPE 0x2A
#define PACKET3_DRAW_INDIRECT_MULTI 0x2C
#define PACKET3_DRAW_INDEX_AUTO 0x2D
#define PACKET3_NUM_INSTANCES 0x2F
#define PACKET3_DRAW_INDEX_MULTI_AUTO 0x30
#define PACKET3_INDIRECT_BUFFER_PRIV 0x32
#define PACKET3_INDIRECT_BUFFER_CNST 0x33
#define PACKET3_COND_INDIRECT_BUFFER_CNST 0x33
#define PACKET3_STRMOUT_BUFFER_UPDATE 0x34
#define PACKET3_DRAW_INDEX_OFFSET_2 0x35
#define PACKET3_DRAW_PREAMBLE 0x36
#define PACKET3_WRITE_DATA 0x37
#define WRITE_DATA_DST_SEL(x) ((x) << 8)
/* 0 - register
* 1 - memory (sync - via GRBM)
* 2 - gl2
* 3 - gds
* 4 - reserved
* 5 - memory (async - direct)
*/
#define WR_ONE_ADDR (1 << 16)
#define WR_CONFIRM (1 << 20)
#define WRITE_DATA_CACHE_POLICY(x) ((x) << 25)
/* 0 - LRU
* 1 - Stream
*/
#define WRITE_DATA_ENGINE_SEL(x) ((x) << 30)
/* 0 - me
* 1 - pfp
* 2 - ce
*/
#define PACKET3_DRAW_INDEX_INDIRECT_MULTI 0x38
#define PACKET3_MEM_SEMAPHORE 0x39
# define PACKET3_SEM_USE_MAILBOX (0x1 << 16)
# define PACKET3_SEM_SEL_SIGNAL_TYPE (0x1 << 20) /* 0 = increment, 1 = write 1 */
# define PACKET3_SEM_SEL_SIGNAL (0x6 << 29)
# define PACKET3_SEM_SEL_WAIT (0x7 << 29)
#define PACKET3_DRAW_INDEX_MULTI_INST 0x3A
#define PACKET3_COPY_DW 0x3B
#define PACKET3_WAIT_REG_MEM 0x3C
#define WAIT_REG_MEM_FUNCTION(x) ((x) << 0)
/* 0 - always
* 1 - <
* 2 - <=
* 3 - ==
* 4 - !=
* 5 - >=
* 6 - >
*/
#define WAIT_REG_MEM_MEM_SPACE(x) ((x) << 4)
/* 0 - reg
* 1 - mem
*/
#define WAIT_REG_MEM_OPERATION(x) ((x) << 6)
/* 0 - wait_reg_mem
* 1 - wr_wait_wr_reg
*/
#define WAIT_REG_MEM_ENGINE(x) ((x) << 8)
/* 0 - me
* 1 - pfp
*/
#define PACKET3_INDIRECT_BUFFER 0x3F
#define INDIRECT_BUFFER_VALID (1 << 23)
#define INDIRECT_BUFFER_CACHE_POLICY(x) ((x) << 28)
/* 0 - LRU
* 1 - Stream
* 2 - Bypass
*/
#define INDIRECT_BUFFER_PRE_ENB(x) ((x) << 21)
#define INDIRECT_BUFFER_PRE_RESUME(x) ((x) << 30)
#define PACKET3_COND_INDIRECT_BUFFER 0x3F
#define PACKET3_COPY_DATA 0x40
#define PACKET3_CP_DMA 0x41
#define PACKET3_PFP_SYNC_ME 0x42
#define PACKET3_SURFACE_SYNC 0x43
#define PACKET3_ME_INITIALIZE 0x44
#define PACKET3_COND_WRITE 0x45
#define PACKET3_EVENT_WRITE 0x46
#define EVENT_TYPE(x) ((x) << 0)
#define EVENT_INDEX(x) ((x) << 8)
/* 0 - any non-TS event
* 1 - ZPASS_DONE, PIXEL_PIPE_STAT_*
* 2 - SAMPLE_PIPELINESTAT
* 3 - SAMPLE_STREAMOUTSTAT*
* 4 - *S_PARTIAL_FLUSH
*/
#define PACKET3_EVENT_WRITE_EOP 0x47
#define PACKET3_EVENT_WRITE_EOS 0x48
#define PACKET3_RELEASE_MEM 0x49
#define PACKET3_RELEASE_MEM_EVENT_TYPE(x) ((x) << 0)
#define PACKET3_RELEASE_MEM_EVENT_INDEX(x) ((x) << 8)
#define PACKET3_RELEASE_MEM_GCR_GLM_WB (1 << 12)
#define PACKET3_RELEASE_MEM_GCR_GLM_INV (1 << 13)
#define PACKET3_RELEASE_MEM_GCR_GLV_INV (1 << 14)
#define PACKET3_RELEASE_MEM_GCR_GL1_INV (1 << 15)
#define PACKET3_RELEASE_MEM_GCR_GL2_US (1 << 16)
#define PACKET3_RELEASE_MEM_GCR_GL2_RANGE (1 << 17)
#define PACKET3_RELEASE_MEM_GCR_GL2_DISCARD (1 << 19)
#define PACKET3_RELEASE_MEM_GCR_GL2_INV (1 << 20)
#define PACKET3_RELEASE_MEM_GCR_GL2_WB (1 << 21)
#define PACKET3_RELEASE_MEM_GCR_SEQ (1 << 22)
#define PACKET3_RELEASE_MEM_CACHE_POLICY(x) ((x) << 25)
/* 0 - cache_policy__me_release_mem__lru
* 1 - cache_policy__me_release_mem__stream
* 2 - cache_policy__me_release_mem__noa
* 3 - cache_policy__me_release_mem__bypass
*/
#define PACKET3_RELEASE_MEM_EXECUTE (1 << 28)
#define PACKET3_RELEASE_MEM_DATA_SEL(x) ((x) << 29)
/* 0 - discard
* 1 - send low 32bit data
* 2 - send 64bit data
* 3 - send 64bit GPU counter value
* 4 - send 64bit sys counter value
*/
#define PACKET3_RELEASE_MEM_INT_SEL(x) ((x) << 24)
/* 0 - none
* 1 - interrupt only (DATA_SEL = 0)
* 2 - interrupt when data write is confirmed
*/
#define PACKET3_RELEASE_MEM_DST_SEL(x) ((x) << 16)
/* 0 - MC
* 1 - TC/L2
*/
#define PACKET3_PREAMBLE_CNTL 0x4A
# define PACKET3_PREAMBLE_BEGIN_CLEAR_STATE (2 << 28)
# define PACKET3_PREAMBLE_END_CLEAR_STATE (3 << 28)
#define PACKET3_DMA_DATA 0x50
/* 1. header
* 2. CONTROL
* 3. SRC_ADDR_LO or DATA [31:0]
* 4. SRC_ADDR_HI [31:0]
* 5. DST_ADDR_LO [31:0]
* 6. DST_ADDR_HI [7:0]
* 7. COMMAND [31:26] | BYTE_COUNT [25:0]
*/
/* CONTROL */
# define PACKET3_DMA_DATA_ENGINE(x) ((x) << 0)
/* 0 - ME
* 1 - PFP
*/
# define PACKET3_DMA_DATA_SRC_CACHE_POLICY(x) ((x) << 13)
/* 0 - LRU
* 1 - Stream
*/
# define PACKET3_DMA_DATA_DST_SEL(x) ((x) << 20)
/* 0 - DST_ADDR using DAS
* 1 - GDS
* 3 - DST_ADDR using L2
*/
# define PACKET3_DMA_DATA_DST_CACHE_POLICY(x) ((x) << 25)
/* 0 - LRU
* 1 - Stream
*/
# define PACKET3_DMA_DATA_SRC_SEL(x) ((x) << 29)
/* 0 - SRC_ADDR using SAS
* 1 - GDS
* 2 - DATA
* 3 - SRC_ADDR using L2
*/
# define PACKET3_DMA_DATA_CP_SYNC (1 << 31)
/* COMMAND */
# define PACKET3_DMA_DATA_CMD_SAS (1 << 26)
/* 0 - memory
* 1 - register
*/
# define PACKET3_DMA_DATA_CMD_DAS (1 << 27)
/* 0 - memory
* 1 - register
*/
# define PACKET3_DMA_DATA_CMD_SAIC (1 << 28)
# define PACKET3_DMA_DATA_CMD_DAIC (1 << 29)
# define PACKET3_DMA_DATA_CMD_RAW_WAIT (1 << 30)
#define PACKET3_CONTEXT_REG_RMW 0x51
#define PACKET3_GFX_CNTX_UPDATE 0x52
#define PACKET3_BLK_CNTX_UPDATE 0x53
#define PACKET3_INCR_UPDT_STATE 0x55
#define PACKET3_ACQUIRE_MEM 0x58
#define PACKET3_REWIND 0x59
#define PACKET3_INTERRUPT 0x5A
#define PACKET3_GEN_PDEPTE 0x5B
#define PACKET3_INDIRECT_BUFFER_PASID 0x5C
#define PACKET3_PRIME_UTCL2 0x5D
#define PACKET3_LOAD_UCONFIG_REG 0x5E
#define PACKET3_LOAD_SH_REG 0x5F
#define PACKET3_LOAD_CONFIG_REG 0x60
#define PACKET3_LOAD_CONTEXT_REG 0x61
#define PACKET3_LOAD_COMPUTE_STATE 0x62
#define PACKET3_LOAD_SH_REG_INDEX 0x63
#define PACKET3_SET_CONFIG_REG 0x68
#define PACKET3_SET_CONFIG_REG_START 0x00002000
#define PACKET3_SET_CONFIG_REG_END 0x00002c00
#define PACKET3_SET_CONTEXT_REG 0x69
#define PACKET3_SET_CONTEXT_REG_START 0x0000a000
#define PACKET3_SET_CONTEXT_REG_END 0x0000a400
#define PACKET3_SET_CONTEXT_REG_INDEX 0x6A
#define PACKET3_SET_VGPR_REG_DI_MULTI 0x71
#define PACKET3_SET_SH_REG_DI 0x72
#define PACKET3_SET_CONTEXT_REG_INDIRECT 0x73
#define PACKET3_SET_SH_REG_DI_MULTI 0x74
#define PACKET3_GFX_PIPE_LOCK 0x75
#define PACKET3_SET_SH_REG 0x76
#define PACKET3_SET_SH_REG_START 0x00002c00
#define PACKET3_SET_SH_REG_END 0x00003000
#define PACKET3_SET_SH_REG_OFFSET 0x77
#define PACKET3_SET_QUEUE_REG 0x78
#define PACKET3_SET_UCONFIG_REG 0x79
#define PACKET3_SET_UCONFIG_REG_START 0x0000c000
#define PACKET3_SET_UCONFIG_REG_END 0x0000c400
#define PACKET3_SET_UCONFIG_REG_INDEX 0x7A
#define PACKET3_FORWARD_HEADER 0x7C
#define PACKET3_SCRATCH_RAM_WRITE 0x7D
#define PACKET3_SCRATCH_RAM_READ 0x7E
#define PACKET3_LOAD_CONST_RAM 0x80
#define PACKET3_WRITE_CONST_RAM 0x81
#define PACKET3_DUMP_CONST_RAM 0x83
#define PACKET3_INCREMENT_CE_COUNTER 0x84
#define PACKET3_INCREMENT_DE_COUNTER 0x85
#define PACKET3_WAIT_ON_CE_COUNTER 0x86
#define PACKET3_WAIT_ON_DE_COUNTER_DIFF 0x88
#define PACKET3_SWITCH_BUFFER 0x8B
#define PACKET3_DISPATCH_DRAW_PREAMBLE 0x8C
#define PACKET3_DISPATCH_DRAW_PREAMBLE_ACE 0x8C
#define PACKET3_DISPATCH_DRAW 0x8D
#define PACKET3_DISPATCH_DRAW_ACE 0x8D
#define PACKET3_GET_LOD_STATS 0x8E
#define PACKET3_DRAW_MULTI_PREAMBLE 0x8F
#define PACKET3_FRAME_CONTROL 0x90
# define FRAME_CMD(x) ((x) << 28)
/*
* x=0: tmz_begin
* x=1: tmz_end
*/
#define PACKET3_INDEX_ATTRIBUTES_INDIRECT 0x91
#define PACKET3_WAIT_REG_MEM64 0x93
#define PACKET3_COND_PREEMPT 0x94
#define PACKET3_HDP_FLUSH 0x95
#define PACKET3_COPY_DATA_RB 0x96
#define PACKET3_INVALIDATE_TLBS 0x98
# define PACKET3_INVALIDATE_TLBS_DST_SEL(x) ((x) << 0)
# define PACKET3_INVALIDATE_TLBS_ALL_HUB(x) ((x) << 4)
# define PACKET3_INVALIDATE_TLBS_PASID(x) ((x) << 5)
#define PACKET3_AQL_PACKET 0x99
#define PACKET3_DMA_DATA_FILL_MULTI 0x9A
#define PACKET3_SET_SH_REG_INDEX 0x9B
#define PACKET3_DRAW_INDIRECT_COUNT_MULTI 0x9C
#define PACKET3_DRAW_INDEX_INDIRECT_COUNT_MULTI 0x9D
#define PACKET3_DUMP_CONST_RAM_OFFSET 0x9E
#define PACKET3_LOAD_CONTEXT_REG_INDEX 0x9F
#define PACKET3_SET_RESOURCES 0xA0
/* 1. header
* 2. CONTROL
* 3. QUEUE_MASK_LO [31:0]
* 4. QUEUE_MASK_HI [31:0]
* 5. GWS_MASK_LO [31:0]
* 6. GWS_MASK_HI [31:0]
* 7. OAC_MASK [15:0]
* 8. GDS_HEAP_SIZE [16:11] | GDS_HEAP_BASE [5:0]
*/
# define PACKET3_SET_RESOURCES_VMID_MASK(x) ((x) << 0)
# define PACKET3_SET_RESOURCES_UNMAP_LATENTY(x) ((x) << 16)
# define PACKET3_SET_RESOURCES_QUEUE_TYPE(x) ((x) << 29)
#define PACKET3_MAP_PROCESS 0xA1
#define PACKET3_MAP_QUEUES 0xA2
/* 1. header
* 2. CONTROL
* 3. CONTROL2
* 4. MQD_ADDR_LO [31:0]
* 5. MQD_ADDR_HI [31:0]
* 6. WPTR_ADDR_LO [31:0]
* 7. WPTR_ADDR_HI [31:0]
*/
/* CONTROL */
# define PACKET3_MAP_QUEUES_QUEUE_SEL(x) ((x) << 4)
# define PACKET3_MAP_QUEUES_VMID(x) ((x) << 8)
# define PACKET3_MAP_QUEUES_QUEUE(x) ((x) << 13)
# define PACKET3_MAP_QUEUES_PIPE(x) ((x) << 16)
# define PACKET3_MAP_QUEUES_ME(x) ((x) << 18)
# define PACKET3_MAP_QUEUES_QUEUE_TYPE(x) ((x) << 21)
# define PACKET3_MAP_QUEUES_ALLOC_FORMAT(x) ((x) << 24)
# define PACKET3_MAP_QUEUES_ENGINE_SEL(x) ((x) << 26)
# define PACKET3_MAP_QUEUES_NUM_QUEUES(x) ((x) << 29)
/* CONTROL2 */
# define PACKET3_MAP_QUEUES_CHECK_DISABLE(x) ((x) << 1)
# define PACKET3_MAP_QUEUES_DOORBELL_OFFSET(x) ((x) << 2)
#define PACKET3_UNMAP_QUEUES 0xA3
/* 1. header
* 2. CONTROL
* 3. CONTROL2
* 4. CONTROL3
* 5. CONTROL4
* 6. CONTROL5
*/
/* CONTROL */
# define PACKET3_UNMAP_QUEUES_ACTION(x) ((x) << 0)
/* 0 - PREEMPT_QUEUES
* 1 - RESET_QUEUES
* 2 - DISABLE_PROCESS_QUEUES
* 3 - PREEMPT_QUEUES_NO_UNMAP
*/
# define PACKET3_UNMAP_QUEUES_QUEUE_SEL(x) ((x) << 4)
# define PACKET3_UNMAP_QUEUES_ENGINE_SEL(x) ((x) << 26)
# define PACKET3_UNMAP_QUEUES_NUM_QUEUES(x) ((x) << 29)
/* CONTROL2a */
# define PACKET3_UNMAP_QUEUES_PASID(x) ((x) << 0)
/* CONTROL2b */
# define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET0(x) ((x) << 2)
/* CONTROL3a */
# define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET1(x) ((x) << 2)
/* CONTROL3b */
# define PACKET3_UNMAP_QUEUES_RB_WPTR(x) ((x) << 0)
/* CONTROL4 */
# define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET2(x) ((x) << 2)
/* CONTROL5 */
# define PACKET3_UNMAP_QUEUES_DOORBELL_OFFSET3(x) ((x) << 2)
#define PACKET3_QUERY_STATUS 0xA4
/* 1. header
* 2. CONTROL
* 3. CONTROL2
* 4. ADDR_LO [31:0]
* 5. ADDR_HI [31:0]
* 6. DATA_LO [31:0]
* 7. DATA_HI [31:0]
*/
/* CONTROL */
# define PACKET3_QUERY_STATUS_CONTEXT_ID(x) ((x) << 0)
# define PACKET3_QUERY_STATUS_INTERRUPT_SEL(x) ((x) << 28)
# define PACKET3_QUERY_STATUS_COMMAND(x) ((x) << 30)
/* CONTROL2a */
# define PACKET3_QUERY_STATUS_PASID(x) ((x) << 0)
/* CONTROL2b */
# define PACKET3_QUERY_STATUS_DOORBELL_OFFSET(x) ((x) << 2)
# define PACKET3_QUERY_STATUS_ENG_SEL(x) ((x) << 25)
#define PACKET3_RUN_LIST 0xA5
#define PACKET3_MAP_PROCESS_VM 0xA6
#endif

118
drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
Просмотреть файл

@ -80,6 +80,18 @@ struct psp_gfx_ctrl
*/
#define GFX_FLAG_RESPONSE 0x80000000
/* Gbr IH registers ID */
enum ih_reg_id {
IH_RB = 0, // IH_RB_CNTL
IH_RB_RNG1 = 1, // IH_RB_CNTL_RING1
IH_RB_RNG2 = 2, // IH_RB_CNTL_RING2
};
/* Command to setup Gibraltar IH register */
struct psp_gfx_cmd_gbr_ih_reg {
uint32_t reg_value; /* Value to be set to the IH_RB_CNTL... register*/
enum ih_reg_id reg_id; /* ID of the register */
};
/* TEE Gfx Command IDs for the ring buffer interface. */
enum psp_gfx_cmd_id
@ -95,9 +107,11 @@ enum psp_gfx_cmd_id
GFX_CMD_ID_SETUP_VMR = 0x00000009, /* setup VMR region */
GFX_CMD_ID_DESTROY_VMR = 0x0000000A, /* destroy VMR region */
GFX_CMD_ID_PROG_REG = 0x0000000B, /* program regs */
/* IDs upto 0x1F are reserved for older programs (Raven, Vega 10/12/20) */
GFX_CMD_ID_LOAD_TOC = 0x00000020, /* Load TOC and obtain TMR size */
GFX_CMD_ID_AUTOLOAD_RLC = 0x00000021, /* Indicates all graphics fw loaded, start RLC autoload */
};
/* Command to load Trusted Application binary into PSP OS. */
struct psp_gfx_cmd_load_ta
{
@ -169,33 +183,59 @@ struct psp_gfx_cmd_setup_tmr
/* FW types for GFX_CMD_ID_LOAD_IP_FW command. Limit 31. */
enum psp_gfx_fw_type
{
GFX_FW_TYPE_NONE = 0,
GFX_FW_TYPE_CP_ME = 1,
GFX_FW_TYPE_CP_PFP = 2,
GFX_FW_TYPE_CP_CE = 3,
GFX_FW_TYPE_CP_MEC = 4,
GFX_FW_TYPE_CP_MEC_ME1 = 5,
GFX_FW_TYPE_CP_MEC_ME2 = 6,
GFX_FW_TYPE_RLC_V = 7,
GFX_FW_TYPE_RLC_G = 8,
GFX_FW_TYPE_SDMA0 = 9,
GFX_FW_TYPE_SDMA1 = 10,
GFX_FW_TYPE_DMCU_ERAM = 11,
GFX_FW_TYPE_DMCU_ISR = 12,
GFX_FW_TYPE_VCN = 13,
GFX_FW_TYPE_UVD = 14,
GFX_FW_TYPE_VCE = 15,
GFX_FW_TYPE_ISP = 16,
GFX_FW_TYPE_ACP = 17,
GFX_FW_TYPE_SMU = 18,
GFX_FW_TYPE_MMSCH = 19,
GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM = 20,
GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM = 21,
GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_CNTL = 22,
GFX_FW_TYPE_UVD1 = 23,
GFX_FW_TYPE_MAX = 24
enum psp_gfx_fw_type {
GFX_FW_TYPE_NONE = 0, /* */
GFX_FW_TYPE_CP_ME = 1, /* CP-ME VG + RV */
GFX_FW_TYPE_CP_PFP = 2, /* CP-PFP VG + RV */
GFX_FW_TYPE_CP_CE = 3, /* CP-CE VG + RV */
GFX_FW_TYPE_CP_MEC = 4, /* CP-MEC FW VG + RV */
GFX_FW_TYPE_CP_MEC_ME1 = 5, /* CP-MEC Jump Table 1 VG + RV */
GFX_FW_TYPE_CP_MEC_ME2 = 6, /* CP-MEC Jump Table 2 VG */
GFX_FW_TYPE_RLC_V = 7, /* RLC-V VG */
GFX_FW_TYPE_RLC_G = 8, /* RLC-G VG + RV */
GFX_FW_TYPE_SDMA0 = 9, /* SDMA0 VG + RV */
GFX_FW_TYPE_SDMA1 = 10, /* SDMA1 VG */
GFX_FW_TYPE_DMCU_ERAM = 11, /* DMCU-ERAM VG + RV */
GFX_FW_TYPE_DMCU_ISR = 12, /* DMCU-ISR VG + RV */
GFX_FW_TYPE_VCN = 13, /* VCN RV */
GFX_FW_TYPE_UVD = 14, /* UVD VG */
GFX_FW_TYPE_VCE = 15, /* VCE VG */
GFX_FW_TYPE_ISP = 16, /* ISP RV */
GFX_FW_TYPE_ACP = 17, /* ACP RV */
GFX_FW_TYPE_SMU = 18, /* SMU VG */
GFX_FW_TYPE_MMSCH = 19, /* MMSCH VG */
GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM = 20, /* RLC GPM VG + RV */
GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM = 21, /* RLC SRM VG + RV */
GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_CNTL = 22, /* RLC CNTL VG + RV */
GFX_FW_TYPE_UVD1 = 23, /* UVD1 VG-20 */
GFX_FW_TYPE_TOC = 24, /* TOC NV-10 */
GFX_FW_TYPE_RLC_P = 25, /* RLC P NV */
GFX_FW_TYPE_RLX6 = 26, /* RLX6 NV */
GFX_FW_TYPE_GLOBAL_TAP_DELAYS = 27, /* GLOBAL TAP DELAYS NV */
GFX_FW_TYPE_SE0_TAP_DELAYS = 28, /* SE0 TAP DELAYS NV */
GFX_FW_TYPE_SE1_TAP_DELAYS = 29, /* SE1 TAP DELAYS NV */
GFX_FW_TYPE_GLOBAL_SE0_SE1_SKEW_DELAYS = 30, /* GLOBAL SE0/1 SKEW DELAYS NV */
GFX_FW_TYPE_SDMA0_JT = 31, /* SDMA0 JT NV */
GFX_FW_TYPE_SDMA1_JT = 32, /* SDNA1 JT NV */
GFX_FW_TYPE_CP_MES = 33, /* CP MES NV */
GFX_FW_TYPE_MES_STACK = 34, /* MES STACK NV */
GFX_FW_TYPE_RLC_SRM_DRAM_SR = 35, /* RLC SRM DRAM NV */
GFX_FW_TYPE_RLCG_SCRATCH_SR = 36, /* RLCG SCRATCH NV */
GFX_FW_TYPE_RLCP_SCRATCH_SR = 37, /* RLCP SCRATCH NV */
GFX_FW_TYPE_RLCV_SCRATCH_SR = 38, /* RLCV SCRATCH NV */
GFX_FW_TYPE_RLX6_DRAM_SR = 39, /* RLX6 DRAM NV */
GFX_FW_TYPE_SDMA0_PG_CONTEXT = 40, /* SDMA0 PG CONTEXT NV */
GFX_FW_TYPE_SDMA1_PG_CONTEXT = 41, /* SDMA1 PG CONTEXT NV */
GFX_FW_TYPE_GLOBAL_MUX_SELECT_RAM = 42, /* GLOBAL MUX SEL RAM NV */
GFX_FW_TYPE_SE0_MUX_SELECT_RAM = 43, /* SE0 MUX SEL RAM NV */
GFX_FW_TYPE_SE1_MUX_SELECT_RAM = 44, /* SE1 MUX SEL RAM NV */
GFX_FW_TYPE_ACCUM_CTRL_RAM = 45, /* ACCUM CTRL RAM NV */
GFX_FW_TYPE_RLCP_CAM = 46, /* RLCP CAM NV */
GFX_FW_TYPE_RLC_SPP_CAM_EXT = 47, /* RLC SPP CAM EXT NV */
GFX_FW_TYPE_RLX6_DRAM_BOOT = 48, /* RLX6 DRAM BOOT NV */
GFX_FW_TYPE_VCN0_RAM = 49, /* VCN_RAM NV */
GFX_FW_TYPE_VCN1_RAM = 50, /* VCN_RAM NV */
GFX_FW_TYPE_MAX
};
/* Command to load HW IP FW. */
@ -224,6 +264,14 @@ struct psp_gfx_cmd_reg_prog {
uint32_t reg_id;
};
/* Command to load TOC */
struct psp_gfx_cmd_load_toc
{
uint32_t toc_phy_addr_lo; /* bits [31:0] of GPU Virtual address of FW location (must be 4 KB aligned) */
uint32_t toc_phy_addr_hi; /* bits [63:32] of GPU Virtual address of FW location */
uint32_t toc_size; /* FW buffer size in bytes */
};
/* All GFX ring buffer commands. */
union psp_gfx_commands
{
@ -234,21 +282,23 @@ union psp_gfx_commands
struct psp_gfx_cmd_load_ip_fw cmd_load_ip_fw;
struct psp_gfx_cmd_save_restore_ip_fw cmd_save_restore_ip_fw;
struct psp_gfx_cmd_reg_prog cmd_setup_reg_prog;
struct psp_gfx_cmd_setup_tmr cmd_setup_vmr;
struct psp_gfx_cmd_load_toc cmd_load_toc;
};
/* Structure of GFX Response buffer.
* For GPCOM I/F it is part of GFX_CMD_RESP buffer, for RBI
* it is separate buffer.
*/
struct psp_gfx_resp
{
uint32_t status; /* +0 status of command execution */
uint32_t session_id; /* +4 session ID in response to LoadTa command */
uint32_t fw_addr_lo; /* +8 bits [31:0] of FW address within TMR (in response to cmd_load_ip_fw command) */
uint32_t fw_addr_hi; /* +12 bits [63:32] of FW address within TMR (in response to cmd_load_ip_fw command) */
uint32_t status; /* +0 status of command execution */
uint32_t session_id; /* +4 session ID in response to LoadTa command */
uint32_t fw_addr_lo; /* +8 bits [31:0] of FW address within TMR (in response to cmd_load_ip_fw command) */
uint32_t fw_addr_hi; /* +12 bits [63:32] of FW address within TMR (in response to cmd_load_ip_fw command) */
uint32_t tmr_size; /* +16 size of the TMR to be reserved including MM fw and Gfx fw in response to cmd_load_toc command */
uint32_t reserved[4];
uint32_t reserved[3];
/* total 32 bytes */
};

117
drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
Просмотреть файл

@ -41,9 +41,16 @@
MODULE_FIRMWARE("amdgpu/vega20_sos.bin");
MODULE_FIRMWARE("amdgpu/vega20_asd.bin");
MODULE_FIRMWARE("amdgpu/vega20_ta.bin");
MODULE_FIRMWARE("amdgpu/navi10_sos.bin");
MODULE_FIRMWARE("amdgpu/navi10_asd.bin");
/* address block */
#define smnMP1_FIRMWARE_FLAGS 0x3010024
/* navi10 reg offset define */
#define mmRLC_GPM_UCODE_ADDR_NV10 0x5b61
#define mmRLC_GPM_UCODE_DATA_NV10 0x5b62
#define mmSDMA0_UCODE_ADDR_NV10 0x5880
#define mmSDMA0_UCODE_DATA_NV10 0x5881
static int psp_v11_0_init_microcode(struct psp_context *psp)
{
@ -52,6 +59,7 @@ static int psp_v11_0_init_microcode(struct psp_context *psp)
char fw_name[30];
int err = 0;
const struct psp_firmware_header_v1_0 *sos_hdr;
const struct psp_firmware_header_v1_1 *sos_hdr_v1_1;
const struct psp_firmware_header_v1_0 *asd_hdr;
const struct ta_firmware_header_v1_0 *ta_hdr;
@ -61,6 +69,9 @@ static int psp_v11_0_init_microcode(struct psp_context *psp)
case CHIP_VEGA20:
chip_name = "vega20";
break;
case CHIP_NAVI10:
chip_name = "navi10";
break;
default:
BUG();
}
@ -75,15 +86,31 @@ static int psp_v11_0_init_microcode(struct psp_context *psp)
goto out;
sos_hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.sos_fw->data;
adev->psp.sos_fw_version = le32_to_cpu(sos_hdr->header.ucode_version);
adev->psp.sos_feature_version = le32_to_cpu(sos_hdr->ucode_feature_version);
adev->psp.sos_bin_size = le32_to_cpu(sos_hdr->sos_size_bytes);
adev->psp.sys_bin_size = le32_to_cpu(sos_hdr->header.ucode_size_bytes) -
le32_to_cpu(sos_hdr->sos_size_bytes);
adev->psp.sys_start_addr = (uint8_t *)sos_hdr +
amdgpu_ucode_print_psp_hdr(&sos_hdr->header);
switch (sos_hdr->header.header_version_major) {
case 1:
adev->psp.sos_fw_version = le32_to_cpu(sos_hdr->header.ucode_version);
adev->psp.sos_feature_version = le32_to_cpu(sos_hdr->ucode_feature_version);
adev->psp.sos_bin_size = le32_to_cpu(sos_hdr->sos_size_bytes);
adev->psp.sys_bin_size = le32_to_cpu(sos_hdr->sos_offset_bytes);
adev->psp.sys_start_addr = (uint8_t *)sos_hdr +
le32_to_cpu(sos_hdr->header.ucode_array_offset_bytes);
adev->psp.sos_start_addr = (uint8_t *)adev->psp.sys_start_addr +
adev->psp.sos_start_addr = (uint8_t *)adev->psp.sys_start_addr +
le32_to_cpu(sos_hdr->sos_offset_bytes);
if (sos_hdr->header.header_version_minor == 1) {
sos_hdr_v1_1 = (const struct psp_firmware_header_v1_1 *)adev->psp.sos_fw->data;
adev->psp.toc_bin_size = le32_to_cpu(sos_hdr_v1_1->toc_size_bytes);
adev->psp.toc_start_addr = (uint8_t *)adev->psp.sys_start_addr +
le32_to_cpu(sos_hdr_v1_1->toc_offset_bytes);
}
break;
default:
dev_err(adev->dev,
"Unsupported psp sos firmware\n");
err = -EINVAL;
goto out;
}
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_asd.bin", chip_name);
err = request_firmware(&adev->psp.asd_fw, fw_name, adev->dev);
@ -101,30 +128,36 @@ static int psp_v11_0_init_microcode(struct psp_context *psp)
adev->psp.asd_start_addr = (uint8_t *)asd_hdr +
le32_to_cpu(asd_hdr->header.ucode_array_offset_bytes);
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ta.bin", chip_name);
err = request_firmware(&adev->psp.ta_fw, fw_name, adev->dev);
if (err) {
release_firmware(adev->psp.ta_fw);
adev->psp.ta_fw = NULL;
dev_info(adev->dev,
"psp v11.0: Failed to load firmware \"%s\"\n", fw_name);
} else {
err = amdgpu_ucode_validate(adev->psp.ta_fw);
if (err)
goto out2;
switch (adev->asic_type) {
case CHIP_VEGA20:
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_ta.bin", chip_name);
err = request_firmware(&adev->psp.ta_fw, fw_name, adev->dev);
if (err) {
release_firmware(adev->psp.ta_fw);
adev->psp.ta_fw = NULL;
dev_info(adev->dev,
"psp v11.0: Failed to load firmware \"%s\"\n", fw_name);
} else {
err = amdgpu_ucode_validate(adev->psp.ta_fw);
if (err)
goto out2;
ta_hdr = (const struct ta_firmware_header_v1_0 *)adev->psp.ta_fw->data;
adev->psp.ta_xgmi_ucode_version = le32_to_cpu(ta_hdr->ta_xgmi_ucode_version);
adev->psp.ta_xgmi_ucode_size = le32_to_cpu(ta_hdr->ta_xgmi_size_bytes);
adev->psp.ta_xgmi_start_addr = (uint8_t *)ta_hdr +
le32_to_cpu(ta_hdr->header.ucode_array_offset_bytes);
adev->psp.ta_fw_version = le32_to_cpu(ta_hdr->header.ucode_version);
adev->psp.ta_ras_ucode_version = le32_to_cpu(ta_hdr->ta_ras_ucode_version);
adev->psp.ta_ras_ucode_size = le32_to_cpu(ta_hdr->ta_ras_size_bytes);
adev->psp.ta_ras_start_addr = (uint8_t *)adev->psp.ta_xgmi_start_addr +
le32_to_cpu(ta_hdr->ta_ras_offset_bytes);
ta_hdr = (const struct ta_firmware_header_v1_0 *)adev->psp.ta_fw->data;
adev->psp.ta_xgmi_ucode_version = le32_to_cpu(ta_hdr->ta_xgmi_ucode_version);
adev->psp.ta_xgmi_ucode_size = le32_to_cpu(ta_hdr->ta_xgmi_size_bytes);
adev->psp.ta_xgmi_start_addr = (uint8_t *)ta_hdr +
le32_to_cpu(ta_hdr->header.ucode_array_offset_bytes);
adev->psp.ta_fw_version = le32_to_cpu(ta_hdr->header.ucode_version);
adev->psp.ta_ras_ucode_version = le32_to_cpu(ta_hdr->ta_ras_ucode_version);
adev->psp.ta_ras_ucode_size = le32_to_cpu(ta_hdr->ta_ras_size_bytes);
adev->psp.ta_ras_start_addr = (uint8_t *)adev->psp.ta_xgmi_start_addr +
le32_to_cpu(ta_hdr->ta_ras_offset_bytes);
}
break;
case CHIP_NAVI10:
break;
default:
BUG();
}
return 0;
@ -501,14 +534,24 @@ psp_v11_0_sram_map(struct amdgpu_device *adev,
case AMDGPU_UCODE_ID_RLC_G:
*sram_offset = 0x2000;
*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_ADDR);
*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_DATA);
if (adev->asic_type != CHIP_NAVI10) {
*sram_addr_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_ADDR);
*sram_data_reg_offset = SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UCODE_DATA);
} else {
*sram_addr_reg_offset = adev->reg_offset[GC_HWIP][0][1] + mmRLC_GPM_UCODE_ADDR_NV10;
*sram_data_reg_offset = adev->reg_offset[GC_HWIP][0][1] + mmRLC_GPM_UCODE_DATA_NV10;
}
break;
case AMDGPU_UCODE_ID_SDMA0:
*sram_offset = 0x0;
*sram_addr_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_ADDR);
*sram_data_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_DATA);
if (adev->asic_type != CHIP_NAVI10) {
*sram_addr_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_ADDR);
*sram_data_reg_offset = SOC15_REG_OFFSET(SDMA0, 0, mmSDMA0_UCODE_DATA);
} else {
*sram_addr_reg_offset = adev->reg_offset[GC_HWIP][0][1] + mmSDMA0_UCODE_ADDR_NV10;
*sram_data_reg_offset = adev->reg_offset[GC_HWIP][0][1] + mmSDMA0_UCODE_DATA_NV10;
}
break;
/* TODO: needs to confirm */
@ -772,6 +815,11 @@ static int psp_v11_0_ras_cure_posion(struct psp_context *psp, uint64_t *mode_ptr
#endif
}
static int psp_v11_0_rlc_autoload_start(struct psp_context *psp)
{
return psp_rlc_autoload_start(psp);
}
static const struct psp_funcs psp_v11_0_funcs = {
.init_microcode = psp_v11_0_init_microcode,
.bootloader_load_sysdrv = psp_v11_0_bootloader_load_sysdrv,
@ -790,6 +838,7 @@ static const struct psp_funcs psp_v11_0_funcs = {
.support_vmr_ring = psp_v11_0_support_vmr_ring,
.ras_trigger_error = psp_v11_0_ras_trigger_error,
.ras_cure_posion = psp_v11_0_ras_cure_posion,
.rlc_autoload_start = psp_v11_0_rlc_autoload_start,
};
void psp_v11_0_set_psp_funcs(struct psp_context *psp)

7
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
Просмотреть файл

@ -1095,7 +1095,7 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
static int sdma_v4_0_start(struct amdgpu_device *adev)
{
struct amdgpu_ring *ring;
int i, r;
int i, r = 0;
if (amdgpu_sriov_vf(adev)) {
sdma_v4_0_ctx_switch_enable(adev, false);
@ -1569,9 +1569,7 @@ static int sdma_v4_0_late_init(void *handle)
if (r)
goto interrupt;
r = amdgpu_ras_debugfs_create(adev, &fs_info);
if (r)
goto debugfs;
amdgpu_ras_debugfs_create(adev, &fs_info);
r = amdgpu_ras_sysfs_create(adev, &fs_info);
if (r)
@ -1592,7 +1590,6 @@ irq:
amdgpu_ras_sysfs_remove(adev, *ras_if);
sysfs:
amdgpu_ras_debugfs_remove(adev, *ras_if);
debugfs:
amdgpu_ras_interrupt_remove_handler(adev, &ih_info);
interrupt:
amdgpu_ras_feature_enable(adev, *ras_if, 0);

1687
drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c Normal file
Просмотреть файл

Разница между файлами не показана из-за своего большого размера Загрузить разницу

45
drivers/gpu/drm/amd/amdgpu/sdma_v5_0.h Normal file
Просмотреть файл

@ -0,0 +1,45 @@
/*
* Copyright 2019 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 __SDMA_V5_0_H__
#define __SDMA_V5_0_H__
enum sdma_v5_0_utcl2_cache_read_policy {
CACHE_READ_POLICY_L2__LRU = 0x00000000,
CACHE_READ_POLICY_L2__STREAM = 0x00000001,
CACHE_READ_POLICY_L2__NOA = 0x00000002,
CACHE_READ_POLICY_L2__DEFAULT = CACHE_READ_POLICY_L2__NOA,
};
enum sdma_v5_0_utcl2_cache_write_policy {
CACHE_WRITE_POLICY_L2__LRU = 0x00000000,
CACHE_WRITE_POLICY_L2__STREAM = 0x00000001,
CACHE_WRITE_POLICY_L2__NOA = 0x00000002,
CACHE_WRITE_POLICY_L2__BYPASS = 0x00000003,
CACHE_WRITE_POLICY_L2__DEFAULT = CACHE_WRITE_POLICY_L2__BYPASS,
};
extern const struct amd_ip_funcs sdma_v5_0_ip_funcs;
extern const struct amdgpu_ip_block_version sdma_v5_0_ip_block;
#endif /* __SDMA_V5_0_H__ */

4
drivers/gpu/drm/amd/amdgpu/si.c
Просмотреть файл

@ -1340,8 +1340,8 @@ static void si_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
/* This reports 0 on APUs, so return to avoid writing/reading registers
* that may or may not be different from their GPU counterparts
*/
if (adev->flags & AMD_IS_APU)
return;
if (adev->flags & AMD_IS_APU)
return;
/* Set the 2 events that we wish to watch, defined above */
/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */

18
drivers/gpu/drm/amd/amdgpu/soc15.c
Просмотреть файл

@ -275,15 +275,6 @@ static bool soc15_read_bios_from_rom(struct amdgpu_device *adev,
return true;
}
struct soc15_allowed_register_entry {
uint32_t hwip;
uint32_t inst;
uint32_t seg;
uint32_t reg_offset;
bool grbm_indexed;
};
static struct soc15_allowed_register_entry soc15_allowed_read_registers[] = {
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS)},
{ SOC15_REG_ENTRY(GC, 0, mmGRBM_STATUS2)},
@ -388,7 +379,7 @@ void soc15_program_register_sequence(struct amdgpu_device *adev,
} else {
tmp = RREG32(reg);
tmp &= ~(entry->and_mask);
tmp |= entry->or_mask;
tmp |= (entry->or_mask & entry->and_mask);
}
if (reg == SOC15_REG_OFFSET(GC, 0, mmPA_SC_BINNER_EVENT_CNTL_3) ||
@ -722,8 +713,8 @@ static void soc15_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
/* This reports 0 on APUs, so return to avoid writing/reading registers
* that may or may not be different from their GPU counterparts
*/
if (adev->flags & AMD_IS_APU)
return;
if (adev->flags & AMD_IS_APU)
return;
/* Set the 2 events that we wish to watch, defined above */
/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
@ -1035,6 +1026,8 @@ static int soc15_common_sw_init(void *handle)
if (amdgpu_sriov_vf(adev))
xgpu_ai_mailbox_add_irq_id(adev);
adev->df_funcs->sw_init(adev);
return 0;
}
@ -1081,6 +1074,7 @@ static int soc15_common_hw_init(void *handle)
*/
if (adev->nbio_funcs->remap_hdp_registers)
adev->nbio_funcs->remap_hdp_registers(adev);
/* enable the doorbell aperture */
soc15_enable_doorbell_aperture(adev, true);
/* HW doorbell routing policy: doorbell writing not

10
drivers/gpu/drm/amd/amdgpu/soc15.h
Просмотреть файл

@ -48,6 +48,16 @@ struct soc15_reg_entry {
uint32_t seg;
uint32_t reg_offset;
uint32_t reg_value;
uint32_t se_num;
uint32_t instance;
};
struct soc15_allowed_register_entry {
uint32_t hwip;
uint32_t inst;
uint32_t seg;
uint32_t reg_offset;
bool grbm_indexed;
};
#define SOC15_REG_ENTRY(ip, inst, reg) ip##_HWIP, inst, reg##_BASE_IDX, reg

108
drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
Просмотреть файл

@ -1,29 +1,26 @@
/****************************************************************************\
*
* File Name ta_ras_if.h
* Project AMD PSP SW IP Module
*
* Description Interface to the RAS Trusted Application
*
* Copyright 2019 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.
*/
/*
* Copyright 2019 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 _TA_RAS_IF_H
#define _TA_RAS_IF_H
@ -31,8 +28,8 @@
#define RSP_ID_MASK (1U << 31)
#define RSP_ID(cmdId) (((uint32_t)(cmdId)) | RSP_ID_MASK)
#define TA_NUM_BLOCK_MAX 14
/* RAS related enumerations */
/**********************************************************/
enum ras_command {
TA_RAS_COMMAND__ENABLE_FEATURES = 0,
TA_RAS_COMMAND__DISABLE_FEATURES,
@ -45,7 +42,12 @@ enum ta_ras_status {
TA_RAS_STATUS__ERROR_INVALID_PARAMETER = 0x02,
TA_RAS_STATUS__ERROR_RAS_NOT_AVAILABLE = 0x03,
TA_RAS_STATUS__ERROR_RAS_DUPLICATE_CMD = 0x04,
TA_RAS_STATUS__ERROR_INJECTION_FAILED = 0x05
TA_RAS_STATUS__ERROR_INJECTION_FAILED = 0x05,
TA_RAS_STATUS__ERROR_ASD_READ_WRITE = 0x06,
TA_RAS_STATUS__ERROR_TOGGLE_DF_CSTATE = 0x07,
TA_RAS_STATUS__ERROR_TIMEOUT = 0x08,
TA_RAS_STATUS__ERROR_BLOCK_DISABLED = 0x09,
TA_RAS_STATUS__ERROR_GENERIC = 0x10,
};
enum ta_ras_block {
@ -62,47 +64,55 @@ enum ta_ras_block {
TA_RAS_BLOCK__SEM,
TA_RAS_BLOCK__MP0,
TA_RAS_BLOCK__MP1,
TA_RAS_BLOCK__FUSE = (TA_NUM_BLOCK_MAX - 1),
TA_RAS_BLOCK__FUSE,
TA_NUM_BLOCK_MAX
};
enum ta_ras_error_type {
TA_RAS_ERROR__NONE = 0,
TA_RAS_ERROR__PARITY = 1,
TA_RAS_ERROR__SINGLE_CORRECTABLE = 2,
TA_RAS_ERROR__MULTI_UNCORRECTABLE = 4,
TA_RAS_ERROR__POISON = 8
TA_RAS_ERROR__NONE = 0,
TA_RAS_ERROR__PARITY = 1,
TA_RAS_ERROR__SINGLE_CORRECTABLE = 2,
TA_RAS_ERROR__MULTI_UNCORRECTABLE = 4,
TA_RAS_ERROR__POISON = 8,
};
/* Input/output structures for RAS commands */
/**********************************************************/
struct ta_ras_enable_features_input {
enum ta_ras_block block_id;
enum ta_ras_error_type error_type;
enum ta_ras_block block_id;
enum ta_ras_error_type error_type;
};
struct ta_ras_disable_features_input {
enum ta_ras_block block_id;
enum ta_ras_error_type error_type;
enum ta_ras_block block_id;
enum ta_ras_error_type error_type;
};
struct ta_ras_trigger_error_input {
enum ta_ras_block block_id;
enum ta_ras_error_type inject_error_type;
uint32_t sub_block_index;
uint64_t address;
uint64_t value;
enum ta_ras_block block_id; // ras-block. i.e. umc, gfx
enum ta_ras_error_type inject_error_type; // type of error. i.e. single_correctable
uint32_t sub_block_index; // mem block. i.e. hbm, sram etc.
uint64_t address; // explicit address of error
uint64_t value; // method if error injection. i.e persistent, coherent etc.
};
/* Common input structure for RAS callbacks */
/**********************************************************/
union ta_ras_cmd_input {
struct ta_ras_enable_features_input enable_features;
struct ta_ras_disable_features_input disable_features;
struct ta_ras_trigger_error_input trigger_error;
};
/* Shared Memory structures */
/**********************************************************/
struct ta_ras_shared_memory {
uint32_t cmd_id;
uint32_t resp_id;
enum ta_ras_status ras_status;
uint32_t reserved;
union ta_ras_cmd_input ras_in_message;
uint32_t cmd_id;
uint32_t resp_id;
enum ta_ras_status ras_status;
uint32_t reserved;
union ta_ras_cmd_input ras_in_message;
};
#endif // TL_RAS_IF_H_

13
drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
Просмотреть файл

@ -128,6 +128,17 @@ static int vcn_v1_0_sw_init(void *handle)
if (r)
return r;
adev->vcn.internal.scratch9 = adev->vcn.external.scratch9 =
SOC15_REG_OFFSET(UVD, 0, mmUVD_SCRATCH9);
adev->vcn.internal.data0 = adev->vcn.external.data0 =
SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA0);
adev->vcn.internal.data1 = adev->vcn.external.data1 =
SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_DATA1);
adev->vcn.internal.cmd = adev->vcn.external.cmd =
SOC15_REG_OFFSET(UVD, 0, mmUVD_GPCOM_VCPU_CMD);
adev->vcn.internal.nop = adev->vcn.external.nop =
SOC15_REG_OFFSET(UVD, 0, mmUVD_NO_OP);
for (i = 0; i < adev->vcn.num_enc_rings; ++i) {
ring = &adev->vcn.ring_enc[i];
sprintf(ring->name, "vcn_enc%d", i);
@ -143,6 +154,8 @@ static int vcn_v1_0_sw_init(void *handle)
return r;
adev->vcn.pause_dpg_mode = vcn_v1_0_pause_dpg_mode;
adev->vcn.internal.jpeg_pitch = adev->vcn.external.jpeg_pitch =
SOC15_REG_OFFSET(UVD, 0, mmUVD_JPEG_PITCH);
return 0;
}

2261
drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c Normal file
Просмотреть файл

Разница между файлами не показана из-за своего большого размера Загрузить разницу

29
drivers/gpu/drm/amd/amdgpu/vcn_v2_0.h Normal file
Просмотреть файл

@ -0,0 +1,29 @@
/*
* Copyright 2018 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 __VCN_V2_0_H__
#define __VCN_V2_0_H__
extern const struct amdgpu_ip_block_version vcn_v2_0_ip_block;
#endif /* __VCN_V2_0_H__ */

3
drivers/gpu/drm/amd/amdkfd/Makefile
Просмотреть файл

@ -36,16 +36,19 @@ AMDKFD_FILES := $(AMDKFD_PATH)/kfd_module.o \
$(AMDKFD_PATH)/kfd_mqd_manager_cik.o \
$(AMDKFD_PATH)/kfd_mqd_manager_vi.o \
$(AMDKFD_PATH)/kfd_mqd_manager_v9.o \
$(AMDKFD_PATH)/kfd_mqd_manager_v10.o \
$(AMDKFD_PATH)/kfd_kernel_queue.o \
$(AMDKFD_PATH)/kfd_kernel_queue_cik.o \
$(AMDKFD_PATH)/kfd_kernel_queue_vi.o \
$(AMDKFD_PATH)/kfd_kernel_queue_v9.o \
$(AMDKFD_PATH)/kfd_kernel_queue_v10.o \
$(AMDKFD_PATH)/kfd_packet_manager.o \
$(AMDKFD_PATH)/kfd_process_queue_manager.o \
$(AMDKFD_PATH)/kfd_device_queue_manager.o \
$(AMDKFD_PATH)/kfd_device_queue_manager_cik.o \
$(AMDKFD_PATH)/kfd_device_queue_manager_vi.o \
$(AMDKFD_PATH)/kfd_device_queue_manager_v9.o \
$(AMDKFD_PATH)/kfd_device_queue_manager_v10.o \
$(AMDKFD_PATH)/kfd_interrupt.o \
$(AMDKFD_PATH)/kfd_events.o \
$(AMDKFD_PATH)/cik_event_interrupt.o \

299
drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h
Просмотреть файл

@ -561,3 +561,302 @@ static const uint32_t cwsr_trap_gfx9_hex[] = {
0xbf8a0000, 0x95806f6c,
0xbf810000, 0x00000000,
};
static const uint32_t cwsr_trap_gfx10_hex[] = {
0xbf820001, 0xbf82012e,
0xb0804004, 0xb970f802,
0x8a708670, 0xb971f803,
0x8771ff71, 0x00000400,
0xbf850008, 0xb971f803,
0x8771ff71, 0x000001ff,
0xbf850001, 0x806c846c,
0x876dff6d, 0x0000ffff,
0xbe80226c, 0xb971f803,
0x8771ff71, 0x00000100,
0xbf840006, 0xbef60380,
0xb9f60203, 0x876dff6d,
0x0000ffff, 0x80ec886c,
0x82ed806d, 0xbef60380,
0xb9f60283, 0xb973f816,
0xb9762c07, 0x8f769c76,
0x886d766d, 0xb97603c7,
0x8f769b76, 0x886d766d,
0xb976f807, 0x8776ff76,
0x00007fff, 0xb9f6f807,
0xbeee037e, 0xbeef037f,
0xbefe0480, 0xbf900004,
0xbf8e0002, 0xbf88fffe,
0xbef4037e, 0x8775ff7f,
0x0000ffff, 0x8875ff75,
0x00040000, 0xbef60380,
0xbef703ff, 0x00807fac,
0x8776ff7f, 0x08000000,
0x90768376, 0x88777677,
0x8776ff7f, 0x70000000,
0x90768176, 0x88777677,
0xbefb037c, 0xbefa0380,
0xb97202dc, 0x8872727f,
0xbefe03c1, 0x877c8172,
0xbf06817c, 0xbf850002,
0xbeff0380, 0xbf820001,
0xbeff03c1, 0xb9712a05,
0x80718171, 0x8f718271,
0x877c8172, 0xbf06817c,
0xbf85000d, 0x8f768771,
0xbef603ff, 0x01000000,
0xbefc0380, 0x7e008700,
0xe0704000, 0x7a5d0000,
0x807c817c, 0x807aff7a,
0x00000080, 0xbf0a717c,
0xbf85fff8, 0xbf82001b,
0x8f768871, 0xbef603ff,
0x01000000, 0xbefc0380,
0x7e008700, 0xe0704000,
0x7a5d0000, 0x807c817c,
0x807aff7a, 0x00000100,
0xbf0a717c, 0xbf85fff8,
0xb9711e06, 0x8771c171,
0xbf84000c, 0x8f718371,
0x80717c71, 0xbefe03c1,
0xbeff0380, 0x7e008700,
0xe0704000, 0x7a5d0000,
0x807c817c, 0x807aff7a,
0x00000080, 0xbf0a717c,
0xbf85fff8, 0xbf8a0000,
0x8776ff72, 0x04000000,
0xbf84002b, 0xbefe03c1,
0x877c8172, 0xbf06817c,
0xbf850002, 0xbeff0380,
0xbf820001, 0xbeff03c1,
0xb9714306, 0x8771c171,
0xbf840021, 0x8f718671,
0x8f718271, 0xbef60371,
0xbef603ff, 0x01000000,
0xd7650000, 0x000100c1,
0xd7660000, 0x000200c1,
0x16000084, 0x877c8172,
0xbf06817c, 0xbefc0380,
0xbf85000a, 0x807cff7c,
0x00000080, 0x807aff7a,
0x00000080, 0xd5250000,
0x0001ff00, 0x00000080,
0xbf0a717c, 0xbf85fff7,
0xbf820009, 0x807cff7c,
0x00000100, 0x807aff7a,
0x00000100, 0xd5250000,
0x0001ff00, 0x00000100,
0xbf0a717c, 0xbf85fff7,
0x877c8172, 0xbf06817c,
0xbf850003, 0x8f7687ff,
0x0000006a, 0xbf820002,
0x8f7688ff, 0x0000006a,
0xbef603ff, 0x01000000,
0x877c8172, 0xbf06817c,
0xbefc0380, 0xbf800000,
0xbf85000b, 0xbe802e00,
0x7e000200, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000080, 0x807c817c,
0xbf0aff7c, 0x0000006a,
0xbf85fff6, 0xbf82000a,
0xbe802e00, 0x7e000200,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000100,
0x807c817c, 0xbf0aff7c,
0x0000006a, 0xbf85fff6,
0xbef60384, 0xbef603ff,
0x01000000, 0x877c8172,
0xbf06817c, 0xbf850030,
0x7e00027b, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000080, 0x7e00026c,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000080,
0x7e00026d, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000080, 0x7e00026e,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000080,
0x7e00026f, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000080, 0x7e000270,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000080,
0xb971f803, 0x7e000271,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000080,
0x7e000273, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000080, 0xb97bf801,
0x7e00027b, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000080, 0xbf82002f,
0x7e00027b, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000100, 0x7e00026c,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000100,
0x7e00026d, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000100, 0x7e00026e,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000100,
0x7e00026f, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000100, 0x7e000270,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000100,
0xb971f803, 0x7e000271,
0xe0704000, 0x7a5d0000,
0x807aff7a, 0x00000100,
0x7e000273, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000100, 0xb97bf801,
0x7e00027b, 0xe0704000,
0x7a5d0000, 0x807aff7a,
0x00000100, 0xbf820119,
0xbef4037e, 0x8775ff7f,
0x0000ffff, 0x8875ff75,
0x00040000, 0xbef60380,
0xbef703ff, 0x00807fac,
0x8772ff7f, 0x08000000,
0x90728372, 0x88777277,
0x8772ff7f, 0x70000000,
0x90728172, 0x88777277,
0xb97902dc, 0x8879797f,
0xbef80380, 0xbefe03c1,
0x877c8179, 0xbf06817c,
0xbf850002, 0xbeff0380,
0xbf820001, 0xbeff03c1,
0xb96f2a05, 0x806f816f,
0x8f6f826f, 0x877c8179,
0xbf06817c, 0xbf850013,
0x8f76876f, 0xbef603ff,
0x01000000, 0xbef20378,
0x8078ff78, 0x00000080,
0xbefc0381, 0xe0304000,
0x785d0000, 0xbf8c3f70,
0x7e008500, 0x807c817c,
0x8078ff78, 0x00000080,
0xbf0a6f7c, 0xbf85fff7,
0xe0304000, 0x725d0000,
0xbf820023, 0x8f76886f,
0xbef603ff, 0x01000000,
0xbef20378, 0x8078ff78,
0x00000100, 0xbefc0381,
0xe0304000, 0x785d0000,
0xbf8c3f70, 0x7e008500,
0x807c817c, 0x8078ff78,
0x00000100, 0xbf0a6f7c,
0xbf85fff7, 0xb96f1e06,
0x876fc16f, 0xbf84000e,
0x8f6f836f, 0x806f7c6f,
0xbefe03c1, 0xbeff0380,
0xe0304000, 0x785d0000,
0xbf8c3f70, 0x7e008500,
0x807c817c, 0x8078ff78,
0x00000080, 0xbf0a6f7c,
0xbf85fff7, 0xbeff03c1,
0xe0304000, 0x725d0000,
0x8772ff79, 0x04000000,
0xbf840020, 0xbefe03c1,
0x877c8179, 0xbf06817c,
0xbf850002, 0xbeff0380,
0xbf820001, 0xbeff03c1,
0xb96f4306, 0x876fc16f,
0xbf840016, 0x8f6f866f,
0x8f6f826f, 0xbef6036f,
0xbef603ff, 0x01000000,
0x877c8172, 0xbf06817c,
0xbefc0380, 0xbf850007,
0x807cff7c, 0x00000080,
0x8078ff78, 0x00000080,
0xbf0a6f7c, 0xbf85fffa,
0xbf820006, 0x807cff7c,
0x00000100, 0x8078ff78,
0x00000100, 0xbf0a6f7c,
0xbf85fffa, 0x877c8179,
0xbf06817c, 0xbf850003,
0x8f7687ff, 0x0000006a,
0xbf820002, 0x8f7688ff,
0x0000006a, 0xbef603ff,
0x01000000, 0x877c8179,
0xbf06817c, 0xbf850012,
0xf4211cba, 0xf0000000,
0x8078ff78, 0x00000080,
0xbefc0381, 0xf421003a,
0xf0000000, 0x8078ff78,
0x00000080, 0xbf8cc07f,
0xbe803000, 0xbf800000,
0x807c817c, 0xbf0aff7c,
0x0000006a, 0xbf85fff5,
0xbe800372, 0xbf820011,
0xf4211cba, 0xf0000000,
0x8078ff78, 0x00000100,
0xbefc0381, 0xf421003a,
0xf0000000, 0x8078ff78,
0x00000100, 0xbf8cc07f,
0xbe803000, 0xbf800000,
0x807c817c, 0xbf0aff7c,
0x0000006a, 0xbf85fff5,
0xbe800372, 0xbef60384,
0xbef603ff, 0x01000000,
0x877c8179, 0xbf06817c,
0xbf850025, 0xf4211bfa,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211b3a,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211b7a,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211eba,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211efa,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211c3a,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211c7a,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211cfa,
0xf0000000, 0x8078ff78,
0x00000080, 0xf4211e7a,
0xf0000000, 0x8078ff78,
0x00000080, 0xbf820024,
0xf4211bfa, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211b3a, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211b7a, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211eba, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211efa, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211c3a, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211c7a, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211cfa, 0xf0000000,
0x8078ff78, 0x00000100,
0xf4211e7a, 0xf0000000,
0x8078ff78, 0x00000100,
0xbf8cc07f, 0x876dff6d,
0x0000ffff, 0xbefc036f,
0xbefe037a, 0xbeff037b,
0x876f71ff, 0x000003ff,
0xb9ef4803, 0xb9f3f816,
0x876f71ff, 0xfffff800,
0x906f8b6f, 0xb9efa2c3,
0xb9f9f801, 0x876fff6d,
0xf0000000, 0x906f9c6f,
0x8f6f906f, 0xbef20380,
0x88726f72, 0x876fff6d,
0x08000000, 0x906f9b6f,
0x8f6f8f6f, 0x88726f72,
0x876fff70, 0x00800000,
0x906f976f, 0xb9f2f807,
0xb9f0f802, 0xbf8a0000,
0xbe80226c, 0xbf810000,
0xbf9f0000, 0xbf9f0000,
0xbf9f0000, 0xbf9f0000,
0xbf9f0000, 0x00000000,
};

1124
drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm Normal file
Просмотреть файл

Разница между файлами не показана из-за своего большого размера Загрузить разницу

Некоторые файлы не были показаны из-за слишком большого количества измененных файлов Показать больше