A-Team_Digital_Solutions/android_kernel_motorola_sm6375
Watch
2
0
Fork
You've already forked android_kernel_motorola_sm6375
 0
mirror of https://github.com/BobTheBlinker/android_kernel_motorola_sm6375.git synced 2026-10-25 22:23:14 -04:00
Code Issues Projects Releases Packages Wiki Activity

Merge airlied/drm-next into drm-misc-next

Browse source 
Archit requested this backmerge to facilitate merging some patches
depending on changes between -rc2 & -rc5

Signed-off-by: Sean Paul <seanpaul@chromium.org>
This commit is contained in:
Sean Paul 2017-08-18 10:52:44 -04:00
commit 0e8841ec7e
1094 changed files with 21968 additions and 23619 deletions
Download patch file Download diff file Expand all files Collapse all files

2
drivers/gpu/drm/amd/amdgpu/Makefile
View file

@ -25,7 +25,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \
amdgpu_prime.o amdgpu_vm.o amdgpu_ib.o amdgpu_pll.o \
amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \
amdgpu_gtt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o amdgpu_atomfirmware.o \
amdgpu_queue_mgr.o
amdgpu_queue_mgr.o amdgpu_vf_error.o
# add asic specific block
amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \

218
drivers/gpu/drm/amd/amdgpu/amdgpu.h
View file

@ -68,13 +68,16 @@
#include "gpu_scheduler.h"
#include "amdgpu_virt.h"
#include "amdgpu_gart.h"
/*
* Modules parameters.
*/
extern int amdgpu_modeset;
extern int amdgpu_vram_limit;
extern int amdgpu_gart_size;
extern int amdgpu_vis_vram_limit;
extern unsigned amdgpu_gart_size;
extern int amdgpu_gtt_size;
extern int amdgpu_moverate;
extern int amdgpu_benchmarking;
extern int amdgpu_testing;
@ -104,6 +107,7 @@ extern unsigned amdgpu_pcie_gen_cap;
extern unsigned amdgpu_pcie_lane_cap;
extern unsigned amdgpu_cg_mask;
extern unsigned amdgpu_pg_mask;
extern unsigned amdgpu_sdma_phase_quantum;
extern char *amdgpu_disable_cu;
extern char *amdgpu_virtual_display;
extern unsigned amdgpu_pp_feature_mask;
@ -369,78 +373,10 @@ struct amdgpu_clock {
};
/*
* BO.
* GEM.
*/
struct amdgpu_bo_list_entry {
struct amdgpu_bo *robj;
struct ttm_validate_buffer tv;
struct amdgpu_bo_va *bo_va;
uint32_t priority;
struct page **user_pages;
int user_invalidated;
};
struct amdgpu_bo_va_mapping {
struct list_head list;
struct rb_node rb;
uint64_t start;
uint64_t last;
uint64_t __subtree_last;
uint64_t offset;
uint64_t flags;
};
/* bo virtual addresses in a specific vm */
struct amdgpu_bo_va {
/* protected by bo being reserved */
struct list_head bo_list;
struct dma_fence *last_pt_update;
unsigned ref_count;
/* protected by vm mutex and spinlock */
struct list_head vm_status;
/* mappings for this bo_va */
struct list_head invalids;
struct list_head valids;
/* constant after initialization */
struct amdgpu_vm *vm;
struct amdgpu_bo *bo;
};
#define AMDGPU_GEM_DOMAIN_MAX 0x3
struct amdgpu_bo {
/* Protected by tbo.reserved */
u32 prefered_domains;
u32 allowed_domains;
struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
struct ttm_placement placement;
struct ttm_buffer_object tbo;
struct ttm_bo_kmap_obj kmap;
u64 flags;
unsigned pin_count;
void *kptr;
u64 tiling_flags;
u64 metadata_flags;
void *metadata;
u32 metadata_size;
unsigned prime_shared_count;
/* list of all virtual address to which this bo
* is associated to
*/
struct list_head va;
/* Constant after initialization */
struct drm_gem_object gem_base;
struct amdgpu_bo *parent;
struct amdgpu_bo *shadow;
struct ttm_bo_kmap_obj dma_buf_vmap;
struct amdgpu_mn *mn;
struct list_head mn_list;
struct list_head shadow_list;
};
#define gem_to_amdgpu_bo(gobj) container_of((gobj), struct amdgpu_bo, gem_base)
void amdgpu_gem_object_free(struct drm_gem_object *obj);
@ -531,49 +467,6 @@ int amdgpu_mode_dumb_mmap(struct drm_file *filp,
int amdgpu_fence_slab_init(void);
void amdgpu_fence_slab_fini(void);
/*
* GART structures, functions & helpers
*/
struct amdgpu_mc;
#define AMDGPU_GPU_PAGE_SIZE 4096
#define AMDGPU_GPU_PAGE_MASK (AMDGPU_GPU_PAGE_SIZE - 1)
#define AMDGPU_GPU_PAGE_SHIFT 12
#define AMDGPU_GPU_PAGE_ALIGN(a) (((a) + AMDGPU_GPU_PAGE_MASK) & ~AMDGPU_GPU_PAGE_MASK)
struct amdgpu_gart {
dma_addr_t table_addr;
struct amdgpu_bo *robj;
void *ptr;
unsigned num_gpu_pages;
unsigned num_cpu_pages;
unsigned table_size;
#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
struct page **pages;
#endif
bool ready;
/* Asic default pte flags */
uint64_t gart_pte_flags;
const struct amdgpu_gart_funcs *gart_funcs;
};
int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev);
void amdgpu_gart_table_ram_free(struct amdgpu_device *adev);
int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev);
void amdgpu_gart_table_vram_free(struct amdgpu_device *adev);
int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev);
void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev);
int amdgpu_gart_init(struct amdgpu_device *adev);
void amdgpu_gart_fini(struct amdgpu_device *adev);
int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
int pages);
int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
int pages, struct page **pagelist,
dma_addr_t *dma_addr, uint64_t flags);
int amdgpu_ttm_recover_gart(struct amdgpu_device *adev);
/*
* VMHUB structures, functions & helpers
*/
@ -598,22 +491,20 @@ struct amdgpu_mc {
* about vram size near mc fb location */
u64 mc_vram_size;
u64 visible_vram_size;
u64 gtt_size;
u64 gtt_start;
u64 gtt_end;
u64 gart_size;
u64 gart_start;
u64 gart_end;
u64 vram_start;
u64 vram_end;
unsigned vram_width;
u64 real_vram_size;
int vram_mtrr;
u64 gtt_base_align;
u64 mc_mask;
const struct firmware *fw; /* MC firmware */
uint32_t fw_version;
struct amdgpu_irq_src vm_fault;
uint32_t vram_type;
uint32_t srbm_soft_reset;
struct amdgpu_mode_mc_save save;
bool prt_warning;
uint64_t stolen_size;
/* apertures */
@ -719,15 +610,15 @@ typedef enum _AMDGPU_DOORBELL64_ASSIGNMENT
/* 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_DOORBELL64_RING0_1 = 0xF8,
AMDGPU_DOORBELL64_RING2_3 = 0xF9,
AMDGPU_DOORBELL64_RING4_5 = 0xFA,
AMDGPU_DOORBELL64_RING6_7 = 0xFB,
AMDGPU_DOORBELL64_UVD_RING0_1 = 0xF8,
AMDGPU_DOORBELL64_UVD_RING2_3 = 0xF9,
AMDGPU_DOORBELL64_UVD_RING4_5 = 0xFA,
AMDGPU_DOORBELL64_UVD_RING6_7 = 0xFB,
AMDGPU_DOORBELL64_UVD_RING0_1 = 0xFC,
AMDGPU_DOORBELL64_UVD_RING2_3 = 0xFD,
AMDGPU_DOORBELL64_UVD_RING4_5 = 0xFE,
AMDGPU_DOORBELL64_UVD_RING6_7 = 0xFF,
AMDGPU_DOORBELL64_VCE_RING0_1 = 0xFC,
AMDGPU_DOORBELL64_VCE_RING2_3 = 0xFD,
AMDGPU_DOORBELL64_VCE_RING4_5 = 0xFE,
AMDGPU_DOORBELL64_VCE_RING6_7 = 0xFF,
AMDGPU_DOORBELL64_MAX_ASSIGNMENT = 0xFF,
AMDGPU_DOORBELL64_INVALID = 0xFFFF
@ -866,6 +757,14 @@ struct amdgpu_fpriv {
/*
* residency list
*/
struct amdgpu_bo_list_entry {
struct amdgpu_bo *robj;
struct ttm_validate_buffer tv;
struct amdgpu_bo_va *bo_va;
uint32_t priority;
struct page **user_pages;
int user_invalidated;
};
struct amdgpu_bo_list {
struct mutex lock;
@ -1159,7 +1058,9 @@ struct amdgpu_cs_parser {
struct list_head validated;
struct dma_fence *fence;
uint64_t bytes_moved_threshold;
uint64_t bytes_moved_vis_threshold;
uint64_t bytes_moved;
uint64_t bytes_moved_vis;
struct amdgpu_bo_list_entry *evictable;
/* user fence */
@ -1230,8 +1131,6 @@ struct amdgpu_wb {
int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb);
void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb);
int amdgpu_wb_get_64bit(struct amdgpu_device *adev, u32 *wb);
void amdgpu_wb_free_64bit(struct amdgpu_device *adev, u32 wb);
void amdgpu_get_pcie_info(struct amdgpu_device *adev);
@ -1525,7 +1424,7 @@ struct amdgpu_device {
bool is_atom_fw;
uint8_t *bios;
uint32_t bios_size;
struct amdgpu_bo *stollen_vga_memory;
struct amdgpu_bo *stolen_vga_memory;
uint32_t bios_scratch_reg_offset;
uint32_t bios_scratch[AMDGPU_BIOS_NUM_SCRATCH];
@ -1557,6 +1456,10 @@ struct amdgpu_device {
spinlock_t gc_cac_idx_lock;
amdgpu_rreg_t gc_cac_rreg;
amdgpu_wreg_t gc_cac_wreg;
/* protects concurrent se_cac register access */
spinlock_t se_cac_idx_lock;
amdgpu_rreg_t se_cac_rreg;
amdgpu_wreg_t se_cac_wreg;
/* protects concurrent ENDPOINT (audio) register access */
spinlock_t audio_endpt_idx_lock;
amdgpu_block_rreg_t audio_endpt_rreg;
@ -1593,6 +1496,7 @@ struct amdgpu_device {
spinlock_t lock;
s64 last_update_us;
s64 accum_us; /* accumulated microseconds */
s64 accum_us_vis; /* for visible VRAM */
u32 log2_max_MBps;
} mm_stats;
@ -1687,6 +1591,8 @@ struct amdgpu_device {
bool has_hw_reset;
u8 reset_magic[AMDGPU_RESET_MAGIC_NUM];
/* record last mm index being written through WREG32*/
unsigned long last_mm_index;
};
static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)
@ -1742,6 +1648,8 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
#define WREG32_DIDT(reg, v) adev->didt_wreg(adev, (reg), (v))
#define RREG32_GC_CAC(reg) adev->gc_cac_rreg(adev, (reg))
#define WREG32_GC_CAC(reg, v) adev->gc_cac_wreg(adev, (reg), (v))
#define RREG32_SE_CAC(reg) adev->se_cac_rreg(adev, (reg))
#define WREG32_SE_CAC(reg, v) adev->se_cac_wreg(adev, (reg), (v))
#define RREG32_AUDIO_ENDPT(block, reg) adev->audio_endpt_rreg(adev, (block), (reg))
#define WREG32_AUDIO_ENDPT(block, reg, v) adev->audio_endpt_wreg(adev, (block), (reg), (v))
#define WREG32_P(reg, val, mask) \
@ -1792,50 +1700,6 @@ void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v);
#define RBIOS16(i) (RBIOS8(i) | (RBIOS8((i)+1) << 8))
#define RBIOS32(i) ((RBIOS16(i)) | (RBIOS16((i)+2) << 16))
/*
* RING helpers.
*/
static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
{
if (ring->count_dw <= 0)
DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
ring->ring[ring->wptr++ & ring->buf_mask] = v;
ring->wptr &= ring->ptr_mask;
ring->count_dw--;
}
static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring, void *src, int count_dw)
{
unsigned occupied, chunk1, chunk2;
void *dst;
if (unlikely(ring->count_dw < count_dw)) {
DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
return;
}
occupied = ring->wptr & ring->buf_mask;
dst = (void *)&ring->ring[occupied];
chunk1 = ring->buf_mask + 1 - occupied;
chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
chunk2 = count_dw - chunk1;
chunk1 <<= 2;
chunk2 <<= 2;
if (chunk1)
memcpy(dst, src, chunk1);
if (chunk2) {
src += chunk1;
dst = (void *)ring->ring;
memcpy(dst, src, chunk2);
}
ring->wptr += count_dw;
ring->wptr &= ring->ptr_mask;
ring->count_dw -= count_dw;
}
static inline struct amdgpu_sdma_instance *
amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
{
@ -1898,7 +1762,6 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
#define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
#define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
#define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
#define amdgpu_display_set_vga_render_state(adev, r) (adev)->mode_info.funcs->set_vga_render_state((adev), (r))
#define amdgpu_display_vblank_get_counter(adev, crtc) (adev)->mode_info.funcs->vblank_get_counter((adev), (crtc))
#define amdgpu_display_vblank_wait(adev, crtc) (adev)->mode_info.funcs->vblank_wait((adev), (crtc))
#define amdgpu_display_backlight_set_level(adev, e, l) (adev)->mode_info.funcs->backlight_set_level((e), (l))
@ -1911,8 +1774,6 @@ amdgpu_get_sdma_instance(struct amdgpu_ring *ring)
#define amdgpu_display_page_flip_get_scanoutpos(adev, crtc, vbl, pos) (adev)->mode_info.funcs->page_flip_get_scanoutpos((adev), (crtc), (vbl), (pos))
#define amdgpu_display_add_encoder(adev, e, s, c) (adev)->mode_info.funcs->add_encoder((adev), (e), (s), (c))
#define amdgpu_display_add_connector(adev, ci, sd, ct, ib, coi, h, r) (adev)->mode_info.funcs->add_connector((adev), (ci), (sd), (ct), (ib), (coi), (h), (r))
#define amdgpu_display_stop_mc_access(adev, s) (adev)->mode_info.funcs->stop_mc_access((adev), (s))
#define amdgpu_display_resume_mc_access(adev, s) (adev)->mode_info.funcs->resume_mc_access((adev), (s))
#define amdgpu_emit_copy_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_copy_buffer((ib), (s), (d), (b))
#define amdgpu_emit_fill_buffer(adev, ib, s, d, b) (adev)->mman.buffer_funcs->emit_fill_buffer((ib), (s), (d), (b))
#define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
@ -1927,7 +1788,8 @@ void amdgpu_pci_config_reset(struct amdgpu_device *adev);
bool amdgpu_need_post(struct amdgpu_device *adev);
void amdgpu_update_display_priority(struct amdgpu_device *adev);
void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes);
void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
u64 num_vis_bytes);
void amdgpu_ttm_placement_from_domain(struct amdgpu_bo *abo, u32 domain);
bool amdgpu_ttm_bo_is_amdgpu_bo(struct ttm_buffer_object *bo);
int amdgpu_ttm_tt_get_user_pages(struct ttm_tt *ttm, struct page **pages);
@ -1943,7 +1805,7 @@ bool amdgpu_ttm_tt_is_readonly(struct ttm_tt *ttm);
uint64_t amdgpu_ttm_tt_pte_flags(struct amdgpu_device *adev, struct ttm_tt *ttm,
struct ttm_mem_reg *mem);
void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64 base);
void amdgpu_gtt_location(struct amdgpu_device *adev, struct amdgpu_mc *mc);
void amdgpu_gart_location(struct amdgpu_device *adev, struct amdgpu_mc *mc);
void amdgpu_ttm_set_active_vram_size(struct amdgpu_device *adev, u64 size);
int amdgpu_ttm_init(struct amdgpu_device *adev);
void amdgpu_ttm_fini(struct amdgpu_device *adev);

72
drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c
View file

@ -285,19 +285,20 @@ static int acp_hw_init(void *handle)
return 0;
else if (r)
return r;
if (adev->asic_type != CHIP_STONEY) {
adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
if (adev->acp.acp_genpd == NULL)
return -ENOMEM;
adev->acp.acp_genpd = kzalloc(sizeof(struct acp_pm_domain), GFP_KERNEL);
if (adev->acp.acp_genpd == NULL)
return -ENOMEM;
adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
adev->acp.acp_genpd->gpd.power_on = acp_poweron;
adev->acp.acp_genpd->gpd.name = "ACP_AUDIO";
adev->acp.acp_genpd->gpd.power_off = acp_poweroff;
adev->acp.acp_genpd->gpd.power_on = acp_poweron;
adev->acp.acp_genpd->cgs_dev = adev->acp.cgs_device;
adev->acp.acp_genpd->cgs_dev = adev->acp.cgs_device;
pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
pm_genpd_init(&adev->acp.acp_genpd->gpd, NULL, false);
}
adev->acp.acp_cell = kzalloc(sizeof(struct mfd_cell) * ACP_DEVS,
GFP_KERNEL);
@ -319,14 +320,29 @@ static int acp_hw_init(void *handle)
return -ENOMEM;
}
i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
switch (adev->asic_type) {
case CHIP_STONEY:
i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
break;
default:
i2s_pdata[0].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET;
}
i2s_pdata[0].cap = DWC_I2S_PLAY;
i2s_pdata[0].snd_rates = SNDRV_PCM_RATE_8000_96000;
i2s_pdata[0].i2s_reg_comp1 = ACP_I2S_COMP1_PLAY_REG_OFFSET;
i2s_pdata[0].i2s_reg_comp2 = ACP_I2S_COMP2_PLAY_REG_OFFSET;
switch (adev->asic_type) {
case CHIP_STONEY:
i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
DW_I2S_QUIRK_COMP_PARAM1 |
DW_I2S_QUIRK_16BIT_IDX_OVERRIDE;
break;
default:
i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
DW_I2S_QUIRK_COMP_PARAM1;
}
i2s_pdata[1].quirks = DW_I2S_QUIRK_COMP_REG_OFFSET |
DW_I2S_QUIRK_COMP_PARAM1;
i2s_pdata[1].cap = DWC_I2S_RECORD;
i2s_pdata[1].snd_rates = SNDRV_PCM_RATE_8000_96000;
i2s_pdata[1].i2s_reg_comp1 = ACP_I2S_COMP1_CAP_REG_OFFSET;
@ -373,12 +389,14 @@ static int acp_hw_init(void *handle)
if (r)
return r;
for (i = 0; i < ACP_DEVS ; i++) {
dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
r = pm_genpd_add_device(&adev->acp.acp_genpd->gpd, dev);
if (r) {
dev_err(dev, "Failed to add dev to genpd\n");
return r;
if (adev->asic_type != CHIP_STONEY) {
for (i = 0; i < ACP_DEVS ; i++) {
dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
r = pm_genpd_add_device(&adev->acp.acp_genpd->gpd, dev);
if (r) {
dev_err(dev, "Failed to add dev to genpd\n");
return r;
}
}
}
@ -398,20 +416,22 @@ static int acp_hw_fini(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
/* return early if no ACP */
if (!adev->acp.acp_genpd)
if (!adev->acp.acp_cell)
return 0;
for (i = 0; i < ACP_DEVS ; i++) {
dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
ret = pm_genpd_remove_device(&adev->acp.acp_genpd->gpd, dev);
/* If removal fails, dont giveup and try rest */
if (ret)
dev_err(dev, "remove dev from genpd failed\n");
if (adev->acp.acp_genpd) {
for (i = 0; i < ACP_DEVS ; i++) {
dev = get_mfd_cell_dev(adev->acp.acp_cell[i].name, i);
ret = pm_genpd_remove_device(&adev->acp.acp_genpd->gpd, dev);
/* If removal fails, dont giveup and try rest */
if (ret)
dev_err(dev, "remove dev from genpd failed\n");
}
kfree(adev->acp.acp_genpd);
}
mfd_remove_devices(adev->acp.parent);
kfree(adev->acp.acp_res);
kfree(adev->acp.acp_genpd);
kfree(adev->acp.acp_cell);
return 0;

4
drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
View file

@ -30,10 +30,10 @@
#include <drm/drmP.h>
#include <drm/drm_crtc_helper.h>
#include "amdgpu.h"
#include "amdgpu_pm.h"
#include "amd_acpi.h"
#include "atom.h"
extern void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
/* Call the ATIF method
*/
/**
@ -289,7 +289,7 @@ out:
* handles it.
* Returns NOTIFY code
*/
int amdgpu_atif_handler(struct amdgpu_device *adev,
static int amdgpu_atif_handler(struct amdgpu_device *adev,
struct acpi_bus_event *event)
{
struct amdgpu_atif *atif = &adev->atif;

46
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
View file

@ -27,7 +27,6 @@
#include "amdgpu_gfx.h"
#include <linux/module.h>
const struct kfd2kgd_calls *kfd2kgd;
const struct kgd2kfd_calls *kgd2kfd;
bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
@ -61,24 +60,6 @@ int amdgpu_amdkfd_init(void)
return ret;
}
bool amdgpu_amdkfd_load_interface(struct amdgpu_device *adev)
{
switch (adev->asic_type) {
#ifdef CONFIG_DRM_AMDGPU_CIK
case CHIP_KAVERI:
kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
break;
#endif
case CHIP_CARRIZO:
kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
break;
default:
return false;
}
return true;
}
void amdgpu_amdkfd_fini(void)
{
if (kgd2kfd) {
@ -89,9 +70,27 @@ void amdgpu_amdkfd_fini(void)
void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
{
if (kgd2kfd)
adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
adev->pdev, kfd2kgd);
const struct kfd2kgd_calls *kfd2kgd;
if (!kgd2kfd)
return;
switch (adev->asic_type) {
#ifdef CONFIG_DRM_AMDGPU_CIK
case CHIP_KAVERI:
kfd2kgd = amdgpu_amdkfd_gfx_7_get_functions();
break;
#endif
case CHIP_CARRIZO:
kfd2kgd = amdgpu_amdkfd_gfx_8_0_get_functions();
break;
default:
dev_info(adev->dev, "kfd not supported on this ASIC\n");
return;
}
adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
adev->pdev, kfd2kgd);
}
void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
@ -184,7 +183,8 @@ int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
return -ENOMEM;
r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, &(*mem)->bo);
AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, NULL, 0,
&(*mem)->bo);
if (r) {
dev_err(adev->dev,
"failed to allocate BO for amdkfd (%d)\n", r);

2
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
View file

@ -39,8 +39,6 @@ struct kgd_mem {
int amdgpu_amdkfd_init(void);
void amdgpu_amdkfd_fini(void);
bool amdgpu_amdkfd_load_interface(struct amdgpu_device *adev);
void amdgpu_amdkfd_suspend(struct amdgpu_device *adev);
int amdgpu_amdkfd_resume(struct amdgpu_device *adev);
void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,

32
drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.c
View file

@ -1686,7 +1686,7 @@ void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock)
{
uint32_t bios_6_scratch;
bios_6_scratch = RREG32(mmBIOS_SCRATCH_6);
bios_6_scratch = RREG32(adev->bios_scratch_reg_offset + 6);
if (lock) {
bios_6_scratch |= ATOM_S6_CRITICAL_STATE;
@ -1696,15 +1696,17 @@ void amdgpu_atombios_scratch_regs_lock(struct amdgpu_device *adev, bool lock)
bios_6_scratch |= ATOM_S6_ACC_MODE;
}
WREG32(mmBIOS_SCRATCH_6, bios_6_scratch);
WREG32(adev->bios_scratch_reg_offset + 6, bios_6_scratch);
}
void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
{
uint32_t bios_2_scratch, bios_6_scratch;
bios_2_scratch = RREG32(mmBIOS_SCRATCH_2);
bios_6_scratch = RREG32(mmBIOS_SCRATCH_6);
adev->bios_scratch_reg_offset = mmBIOS_SCRATCH_0;
bios_2_scratch = RREG32(adev->bios_scratch_reg_offset + 2);
bios_6_scratch = RREG32(adev->bios_scratch_reg_offset + 6);
/* let the bios control the backlight */
bios_2_scratch &= ~ATOM_S2_VRI_BRIGHT_ENABLE;
@ -1715,8 +1717,8 @@ void amdgpu_atombios_scratch_regs_init(struct amdgpu_device *adev)
/* clear the vbios dpms state */
bios_2_scratch &= ~ATOM_S2_DEVICE_DPMS_STATE;
WREG32(mmBIOS_SCRATCH_2, bios_2_scratch);
WREG32(mmBIOS_SCRATCH_6, bios_6_scratch);
WREG32(adev->bios_scratch_reg_offset + 2, bios_2_scratch);
WREG32(adev->bios_scratch_reg_offset + 6, bios_6_scratch);
}
void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev)
@ -1724,7 +1726,7 @@ void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev)
int i;
for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
adev->bios_scratch[i] = RREG32(mmBIOS_SCRATCH_0 + i);
adev->bios_scratch[i] = RREG32(adev->bios_scratch_reg_offset + i);
}
void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev)
@ -1738,20 +1740,30 @@ void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev)
adev->bios_scratch[7] &= ~ATOM_S7_ASIC_INIT_COMPLETE_MASK;
for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
WREG32(mmBIOS_SCRATCH_0 + i, adev->bios_scratch[i]);
WREG32(adev->bios_scratch_reg_offset + i, adev->bios_scratch[i]);
}
void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
bool hung)
{
u32 tmp = RREG32(mmBIOS_SCRATCH_3);
u32 tmp = RREG32(adev->bios_scratch_reg_offset + 3);
if (hung)
tmp |= ATOM_S3_ASIC_GUI_ENGINE_HUNG;
else
tmp &= ~ATOM_S3_ASIC_GUI_ENGINE_HUNG;
WREG32(mmBIOS_SCRATCH_3, tmp);
WREG32(adev->bios_scratch_reg_offset + 3, tmp);
}
bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev)
{
u32 tmp = RREG32(adev->bios_scratch_reg_offset + 7);
if (tmp & ATOM_S7_ASIC_INIT_COMPLETE_MASK)
return false;
else
return true;
}
/* Atom needs data in little endian format

1
drivers/gpu/drm/amd/amdgpu/amdgpu_atombios.h
View file

@ -200,6 +200,7 @@ void amdgpu_atombios_scratch_regs_save(struct amdgpu_device *adev);
void amdgpu_atombios_scratch_regs_restore(struct amdgpu_device *adev);
void amdgpu_atombios_scratch_regs_engine_hung(struct amdgpu_device *adev,
bool hung);
bool amdgpu_atombios_scratch_need_asic_init(struct amdgpu_device *adev);
void amdgpu_atombios_copy_swap(u8 *dst, u8 *src, u8 num_bytes, bool to_le);
int amdgpu_atombios_get_max_vddc(struct amdgpu_device *adev, u8 voltage_type,

161
drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c
View file

@ -66,41 +66,6 @@ void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev)
}
}
void amdgpu_atomfirmware_scratch_regs_save(struct amdgpu_device *adev)
{
int i;
for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
adev->bios_scratch[i] = RREG32(adev->bios_scratch_reg_offset + i);
}
void amdgpu_atomfirmware_scratch_regs_restore(struct amdgpu_device *adev)
{
int i;
/*
* VBIOS will check ASIC_INIT_COMPLETE bit to decide if
* execute ASIC_Init posting via driver
*/
adev->bios_scratch[7] &= ~ATOM_S7_ASIC_INIT_COMPLETE_MASK;
for (i = 0; i < AMDGPU_BIOS_NUM_SCRATCH; i++)
WREG32(adev->bios_scratch_reg_offset + i, adev->bios_scratch[i]);
}
void amdgpu_atomfirmware_scratch_regs_engine_hung(struct amdgpu_device *adev,
bool hung)
{
u32 tmp = RREG32(adev->bios_scratch_reg_offset + 3);
if (hung)
tmp |= ATOM_S3_ASIC_GUI_ENGINE_HUNG;
else
tmp &= ~ATOM_S3_ASIC_GUI_ENGINE_HUNG;
WREG32(adev->bios_scratch_reg_offset + 3, tmp);
}
int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
{
struct atom_context *ctx = adev->mode_info.atom_context;
@ -130,3 +95,129 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev)
ctx->scratch_size_bytes = usage_bytes;
return 0;
}
union igp_info {
struct atom_integrated_system_info_v1_11 v11;
};
/*
* Return vram width from integrated system info table, if available,
* or 0 if not.
*/
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);
u16 data_offset, size;
union igp_info *igp_info;
u8 frev, crev;
/* 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;
}
}
return 0;
}
union firmware_info {
struct atom_firmware_info_v3_1 v31;
};
union smu_info {
struct atom_smu_info_v3_1 v31;
};
union umc_info {
struct atom_umc_info_v3_1 v31;
};
int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev)
{
struct amdgpu_mode_info *mode_info = &adev->mode_info;
struct amdgpu_pll *spll = &adev->clock.spll;
struct amdgpu_pll *mpll = &adev->clock.mpll;
uint8_t frev, crev;
uint16_t data_offset;
int ret = -EINVAL, index;
index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
firmwareinfo);
if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
&frev, &crev, &data_offset)) {
union firmware_info *firmware_info =
(union firmware_info *)(mode_info->atom_context->bios +
data_offset);
adev->clock.default_sclk =
le32_to_cpu(firmware_info->v31.bootup_sclk_in10khz);
adev->clock.default_mclk =
le32_to_cpu(firmware_info->v31.bootup_mclk_in10khz);
adev->pm.current_sclk = adev->clock.default_sclk;
adev->pm.current_mclk = adev->clock.default_mclk;
/* not technically a clock, but... */
adev->mode_info.firmware_flags =
le32_to_cpu(firmware_info->v31.firmware_capability);
ret = 0;
}
index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
smu_info);
if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
&frev, &crev, &data_offset)) {
union smu_info *smu_info =
(union smu_info *)(mode_info->atom_context->bios +
data_offset);
/* system clock */
spll->reference_freq = le32_to_cpu(smu_info->v31.core_refclk_10khz);
spll->reference_div = 0;
spll->min_post_div = 1;
spll->max_post_div = 1;
spll->min_ref_div = 2;
spll->max_ref_div = 0xff;
spll->min_feedback_div = 4;
spll->max_feedback_div = 0xff;
spll->best_vco = 0;
ret = 0;
}
index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
umc_info);
if (amdgpu_atom_parse_data_header(mode_info->atom_context, index, NULL,
&frev, &crev, &data_offset)) {
union umc_info *umc_info =
(union umc_info *)(mode_info->atom_context->bios +
data_offset);
/* memory clock */
mpll->reference_freq = le32_to_cpu(umc_info->v31.mem_refclk_10khz);
mpll->reference_div = 0;
mpll->min_post_div = 1;
mpll->max_post_div = 1;
mpll->min_ref_div = 2;
mpll->max_ref_div = 0xff;
mpll->min_feedback_div = 4;
mpll->max_feedback_div = 0xff;
mpll->best_vco = 0;
ret = 0;
}
return ret;
}

6
drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.h
View file

@ -26,10 +26,8 @@
bool amdgpu_atomfirmware_gpu_supports_virtualization(struct amdgpu_device *adev);
void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev);
void amdgpu_atomfirmware_scratch_regs_save(struct amdgpu_device *adev);
void amdgpu_atomfirmware_scratch_regs_restore(struct amdgpu_device *adev);
void amdgpu_atomfirmware_scratch_regs_engine_hung(struct amdgpu_device *adev,
bool hung);
int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev);
int amdgpu_atomfirmware_get_vram_width(struct amdgpu_device *adev);
int amdgpu_atomfirmware_get_clock_info(struct amdgpu_device *adev);
#endif

6
drivers/gpu/drm/amd/amdgpu/amdgpu_benchmark.c
View file

@ -40,7 +40,7 @@ static int amdgpu_benchmark_do_move(struct amdgpu_device *adev, unsigned size,
for (i = 0; i < n; i++) {
struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
r = amdgpu_copy_buffer(ring, saddr, daddr, size, NULL, &fence,
false);
false, false);
if (r)
goto exit_do_move;
r = dma_fence_wait(fence, false);
@ -81,7 +81,7 @@ static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
n = AMDGPU_BENCHMARK_ITERATIONS;
r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, sdomain, 0, NULL,
NULL, &sobj);
NULL, 0, &sobj);
if (r) {
goto out_cleanup;
}
@ -94,7 +94,7 @@ static void amdgpu_benchmark_move(struct amdgpu_device *adev, unsigned size,
goto out_cleanup;
}
r = amdgpu_bo_create(adev, size, PAGE_SIZE, true, ddomain, 0, NULL,
NULL, &dobj);
NULL, 0, &dobj);
if (r) {
goto out_cleanup;
}

17
drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
View file

@ -86,19 +86,6 @@ static bool check_atom_bios(uint8_t *bios, size_t size)
return false;
}
static bool is_atom_fw(uint8_t *bios)
{
uint16_t bios_header_start = bios[0x48] | (bios[0x49] << 8);
uint8_t frev = bios[bios_header_start + 2];
uint8_t crev = bios[bios_header_start + 3];
if ((frev < 3) ||
((frev == 3) && (crev < 3)))
return false;
return true;
}
/* If you boot an IGP board with a discrete card as the primary,
* the IGP rom is not accessible via the rom bar as the IGP rom is
* part of the system bios. On boot, the system bios puts a
@ -117,7 +104,7 @@ static bool igp_read_bios_from_vram(struct amdgpu_device *adev)
adev->bios = NULL;
vram_base = pci_resource_start(adev->pdev, 0);
bios = ioremap(vram_base, size);
bios = ioremap_wc(vram_base, size);
if (!bios) {
return false;
}
@ -455,6 +442,6 @@ bool amdgpu_get_bios(struct amdgpu_device *adev)
return false;
success:
adev->is_atom_fw = is_atom_fw(adev->bios);
adev->is_atom_fw = (adev->asic_type >= CHIP_VEGA10) ? true : false;
return true;
}

22
drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
View file

@ -83,7 +83,7 @@ static int amdgpu_bo_list_create(struct amdgpu_device *adev,
r = idr_alloc(&fpriv->bo_list_handles, list, 1, 0, GFP_KERNEL);
mutex_unlock(&fpriv->bo_list_lock);
if (r < 0) {
kfree(list);
amdgpu_bo_list_free(list);
return r;
}
*id = r;
@ -136,7 +136,7 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
}
bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
usermm = amdgpu_ttm_tt_get_usermm(bo->tbo.ttm);
if (usermm) {
@ -156,11 +156,11 @@ static int amdgpu_bo_list_set(struct amdgpu_device *adev,
entry->tv.bo = &entry->robj->tbo;
entry->tv.shared = !entry->robj->prime_shared_count;
if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_GDS)
if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GDS)
gds_obj = entry->robj;
if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_GWS)
if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_GWS)
gws_obj = entry->robj;
if (entry->robj->prefered_domains == AMDGPU_GEM_DOMAIN_OA)
if (entry->robj->preferred_domains == AMDGPU_GEM_DOMAIN_OA)
oa_obj = entry->robj;
total_size += amdgpu_bo_size(entry->robj);
@ -198,12 +198,16 @@ amdgpu_bo_list_get(struct amdgpu_fpriv *fpriv, int id)
result = idr_find(&fpriv->bo_list_handles, id);
if (result) {
if (kref_get_unless_zero(&result->refcount))
if (kref_get_unless_zero(&result->refcount)) {
rcu_read_unlock();
mutex_lock(&result->lock);
else
} else {
rcu_read_unlock();
result = NULL;
}
} else {
rcu_read_unlock();
}
rcu_read_unlock();
return result;
}
@ -266,7 +270,7 @@ int amdgpu_bo_list_ioctl(struct drm_device *dev, void *data,
struct amdgpu_fpriv *fpriv = filp->driver_priv;
union drm_amdgpu_bo_list *args = data;
uint32_t handle = args->in.list_handle;
const void __user *uptr = (const void*)(uintptr_t)args->in.bo_info_ptr;
const void __user *uptr = u64_to_user_ptr(args->in.bo_info_ptr);
struct drm_amdgpu_bo_list_entry *info;
struct amdgpu_bo_list *list;

30
drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
View file

@ -124,7 +124,7 @@ static int amdgpu_cgs_alloc_gpu_mem(struct cgs_device *cgs_device,
ret = amdgpu_bo_create_restricted(adev, size, PAGE_SIZE,
true, domain, flags,
NULL, &placement, NULL,
&obj);
0, &obj);
if (ret) {
DRM_ERROR("(%d) bo create failed\n", ret);
return ret;
@ -166,7 +166,7 @@ static int amdgpu_cgs_gmap_gpu_mem(struct cgs_device *cgs_device, cgs_handle_t h
r = amdgpu_bo_reserve(obj, true);
if (unlikely(r != 0))
return r;
r = amdgpu_bo_pin_restricted(obj, obj->prefered_domains,
r = amdgpu_bo_pin_restricted(obj, obj->preferred_domains,
min_offset, max_offset, mcaddr);
amdgpu_bo_unreserve(obj);
return r;
@ -240,6 +240,8 @@ static uint32_t amdgpu_cgs_read_ind_register(struct cgs_device *cgs_device,
return RREG32_DIDT(index);
case CGS_IND_REG_GC_CAC:
return RREG32_GC_CAC(index);
case CGS_IND_REG_SE_CAC:
return RREG32_SE_CAC(index);
case CGS_IND_REG__AUDIO_ENDPT:
DRM_ERROR("audio endpt register access not implemented.\n");
return 0;
@ -266,6 +268,8 @@ static void amdgpu_cgs_write_ind_register(struct cgs_device *cgs_device,
return WREG32_DIDT(index, value);
case CGS_IND_REG_GC_CAC:
return WREG32_GC_CAC(index, value);
case CGS_IND_REG_SE_CAC:
return WREG32_SE_CAC(index, value);
case CGS_IND_REG__AUDIO_ENDPT:
DRM_ERROR("audio endpt register access not implemented.\n");
return;
@ -610,6 +614,17 @@ static int amdgpu_cgs_enter_safe_mode(struct cgs_device *cgs_device,
return 0;
}
static void amdgpu_cgs_lock_grbm_idx(struct cgs_device *cgs_device,
bool lock)
{
CGS_FUNC_ADEV;
if (lock)
mutex_lock(&adev->grbm_idx_mutex);
else
mutex_unlock(&adev->grbm_idx_mutex);
}
static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
enum cgs_ucode_id type,
struct cgs_firmware_info *info)
@ -644,7 +659,7 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
info->version = (uint16_t)le32_to_cpu(header->header.ucode_version);
if (CGS_UCODE_ID_CP_MEC == type)
info->image_size = (header->jt_offset) << 2;
info->image_size = le32_to_cpu(header->jt_offset) << 2;
info->fw_version = amdgpu_get_firmware_version(cgs_device, type);
info->feature_version = (uint16_t)le32_to_cpu(header->ucode_feature_version);
@ -719,7 +734,13 @@ static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
strcpy(fw_name, "amdgpu/polaris12_smc.bin");
break;
case CHIP_VEGA10:
strcpy(fw_name, "amdgpu/vega10_smc.bin");
if ((adev->pdev->device == 0x687f) &&
((adev->pdev->revision == 0xc0) ||
(adev->pdev->revision == 0xc1) ||
(adev->pdev->revision == 0xc3)))
strcpy(fw_name, "amdgpu/vega10_acg_smc.bin");
else
strcpy(fw_name, "amdgpu/vega10_smc.bin");
break;
default:
DRM_ERROR("SMC firmware not supported\n");
@ -1117,6 +1138,7 @@ static const struct cgs_ops amdgpu_cgs_ops = {
.query_system_info = amdgpu_cgs_query_system_info,
.is_virtualization_enabled = amdgpu_cgs_is_virtualization_enabled,
.enter_safe_mode = amdgpu_cgs_enter_safe_mode,
.lock_grbm_idx = amdgpu_cgs_lock_grbm_idx,
};
static const struct cgs_os_ops amdgpu_cgs_os_ops = {

102
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
View file

@ -54,7 +54,7 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
*offset = data->offset;
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
if (amdgpu_ttm_tt_get_usermm(p->uf_entry.robj->tbo.ttm)) {
amdgpu_bo_unref(&p->uf_entry.robj);
@ -90,7 +90,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
}
/* get chunks */
chunk_array_user = (uint64_t __user *)(uintptr_t)(cs->in.chunks);
chunk_array_user = u64_to_user_ptr(cs->in.chunks);
if (copy_from_user(chunk_array, chunk_array_user,
sizeof(uint64_t)*cs->in.num_chunks)) {
ret = -EFAULT;
@ -110,7 +110,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
struct drm_amdgpu_cs_chunk user_chunk;
uint32_t __user *cdata;
chunk_ptr = (void __user *)(uintptr_t)chunk_array[i];
chunk_ptr = u64_to_user_ptr(chunk_array[i]);
if (copy_from_user(&user_chunk, chunk_ptr,
sizeof(struct drm_amdgpu_cs_chunk))) {
ret = -EFAULT;
@ -121,7 +121,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, void *data)
p->chunks[i].length_dw = user_chunk.length_dw;
size = p->chunks[i].length_dw;
cdata = (void __user *)(uintptr_t)user_chunk.chunk_data;
cdata = u64_to_user_ptr(user_chunk.chunk_data);
p->chunks[i].kdata = kvmalloc_array(size, sizeof(uint32_t), GFP_KERNEL);
if (p->chunks[i].kdata == NULL) {
@ -223,10 +223,11 @@ static s64 bytes_to_us(struct amdgpu_device *adev, u64 bytes)
* ticks. The accumulated microseconds (us) are converted to bytes and
* returned.
*/
static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
u64 *max_bytes,
u64 *max_vis_bytes)
{
s64 time_us, increment_us;
u64 max_bytes;
u64 free_vram, total_vram, used_vram;
/* Allow a maximum of 200 accumulated ms. This is basically per-IB
@ -238,8 +239,11 @@ static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
*/
const s64 us_upper_bound = 200000;
if (!adev->mm_stats.log2_max_MBps)
return 0;
if (!adev->mm_stats.log2_max_MBps) {
*max_bytes = 0;
*max_vis_bytes = 0;
return;
}
total_vram = adev->mc.real_vram_size - adev->vram_pin_size;
used_vram = atomic64_read(&adev->vram_usage);
@ -280,23 +284,45 @@ static u64 amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev)
adev->mm_stats.accum_us = max(min_us, adev->mm_stats.accum_us);
}
/* This returns 0 if the driver is in debt to disallow (optional)
/* This is set to 0 if the driver is in debt to disallow (optional)
* buffer moves.
*/
max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
*max_bytes = us_to_bytes(adev, adev->mm_stats.accum_us);
/* Do the same for visible VRAM if half of it is free */
if (adev->mc.visible_vram_size < adev->mc.real_vram_size) {
u64 total_vis_vram = adev->mc.visible_vram_size;
u64 used_vis_vram = atomic64_read(&adev->vram_vis_usage);
if (used_vis_vram < total_vis_vram) {
u64 free_vis_vram = total_vis_vram - used_vis_vram;
adev->mm_stats.accum_us_vis = min(adev->mm_stats.accum_us_vis +
increment_us, us_upper_bound);
if (free_vis_vram >= total_vis_vram / 2)
adev->mm_stats.accum_us_vis =
max(bytes_to_us(adev, free_vis_vram / 2),
adev->mm_stats.accum_us_vis);
}
*max_vis_bytes = us_to_bytes(adev, adev->mm_stats.accum_us_vis);
} else {
*max_vis_bytes = 0;
}
spin_unlock(&adev->mm_stats.lock);
return max_bytes;
}
/* Report how many bytes have really been moved for the last command
* submission. This can result in a debt that can stop buffer migrations
* temporarily.
*/
void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes)
void amdgpu_cs_report_moved_bytes(struct amdgpu_device *adev, u64 num_bytes,
u64 num_vis_bytes)
{
spin_lock(&adev->mm_stats.lock);
adev->mm_stats.accum_us -= bytes_to_us(adev, num_bytes);
adev->mm_stats.accum_us_vis -= bytes_to_us(adev, num_vis_bytes);
spin_unlock(&adev->mm_stats.lock);
}
@ -304,7 +330,7 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
struct amdgpu_bo *bo)
{
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
u64 initial_bytes_moved;
u64 initial_bytes_moved, bytes_moved;
uint32_t domain;
int r;
@ -314,17 +340,35 @@ static int amdgpu_cs_bo_validate(struct amdgpu_cs_parser *p,
/* Don't move this buffer if we have depleted our allowance
* to move it. Don't move anything if the threshold is zero.
*/
if (p->bytes_moved < p->bytes_moved_threshold)
domain = bo->prefered_domains;
else
if (p->bytes_moved < p->bytes_moved_threshold) {
if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
(bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
/* And don't move a CPU_ACCESS_REQUIRED BO to limited
* visible VRAM if we've depleted our allowance to do
* that.
*/
if (p->bytes_moved_vis < p->bytes_moved_vis_threshold)
domain = bo->preferred_domains;
else
domain = bo->allowed_domains;
} else {
domain = bo->preferred_domains;
}
} else {
domain = bo->allowed_domains;
}
retry:
amdgpu_ttm_placement_from_domain(bo, domain);
initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
p->bytes_moved += atomic64_read(&adev->num_bytes_moved) -
initial_bytes_moved;
bytes_moved = atomic64_read(&adev->num_bytes_moved) -
initial_bytes_moved;
p->bytes_moved += bytes_moved;
if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
bo->tbo.mem.mem_type == TTM_PL_VRAM &&
bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT)
p->bytes_moved_vis += bytes_moved;
if (unlikely(r == -ENOMEM) && domain != bo->allowed_domains) {
domain = bo->allowed_domains;
@ -350,7 +394,8 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
struct amdgpu_bo_list_entry *candidate = p->evictable;
struct amdgpu_bo *bo = candidate->robj;
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
u64 initial_bytes_moved;
u64 initial_bytes_moved, bytes_moved;
bool update_bytes_moved_vis;
uint32_t other;
/* If we reached our current BO we can forget it */
@ -370,10 +415,17 @@ static bool amdgpu_cs_try_evict(struct amdgpu_cs_parser *p,
/* Good we can try to move this BO somewhere else */
amdgpu_ttm_placement_from_domain(bo, other);
update_bytes_moved_vis =
adev->mc.visible_vram_size < adev->mc.real_vram_size &&
bo->tbo.mem.mem_type == TTM_PL_VRAM &&
bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT;
initial_bytes_moved = atomic64_read(&adev->num_bytes_moved);
r = ttm_bo_validate(&bo->tbo, &bo->placement, true, false);
p->bytes_moved += atomic64_read(&adev->num_bytes_moved) -
bytes_moved = atomic64_read(&adev->num_bytes_moved) -
initial_bytes_moved;
p->bytes_moved += bytes_moved;
if (update_bytes_moved_vis)
p->bytes_moved_vis += bytes_moved;
if (unlikely(r))
break;
@ -554,8 +606,10 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
list_splice(&need_pages, &p->validated);
}
p->bytes_moved_threshold = amdgpu_cs_get_threshold_for_moves(p->adev);
amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
&p->bytes_moved_vis_threshold);
p->bytes_moved = 0;
p->bytes_moved_vis = 0;
p->evictable = list_last_entry(&p->validated,
struct amdgpu_bo_list_entry,
tv.head);
@ -579,8 +633,8 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
goto error_validate;
}
amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved);
amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
p->bytes_moved_vis);
fpriv->vm.last_eviction_counter =
atomic64_read(&p->adev->num_evictions);
@ -1383,7 +1437,7 @@ int amdgpu_cs_wait_fences_ioctl(struct drm_device *dev, void *data,
if (fences == NULL)
return -ENOMEM;
fences_user = (void __user *)(uintptr_t)(wait->in.fences);
fences_user = u64_to_user_ptr(wait->in.fences);
if (copy_from_user(fences, fences_user,
sizeof(struct drm_amdgpu_fence) * fence_count)) {
r = -EFAULT;

244
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
View file

@ -53,6 +53,9 @@
#include "bif/bif_4_1_d.h"
#include <linux/pci.h>
#include <linux/firmware.h>
#include "amdgpu_vf_error.h"
#include "amdgpu_amdkfd.h"
MODULE_FIRMWARE("amdgpu/vega10_gpu_info.bin");
MODULE_FIRMWARE("amdgpu/raven_gpu_info.bin");
@ -128,6 +131,10 @@ void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
{
trace_amdgpu_mm_wreg(adev->pdev->device, reg, v);
if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
adev->last_mm_index = v;
}
if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev)) {
BUG_ON(in_interrupt());
return amdgpu_virt_kiq_wreg(adev, reg, v);
@ -143,6 +150,10 @@ void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
writel(v, ((void __iomem *)adev->rmmio) + (mmMM_DATA * 4));
spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
}
if (adev->asic_type >= CHIP_VEGA10 && reg == 1 && adev->last_mm_index == 0x5702C) {
udelay(500);
}
}
u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
@ -157,6 +168,9 @@ u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
{
if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
adev->last_mm_index = v;
}
if ((reg * 4) < adev->rio_mem_size)
iowrite32(v, adev->rio_mem + (reg * 4));
@ -164,6 +178,10 @@ void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
iowrite32((reg * 4), adev->rio_mem + (mmMM_INDEX * 4));
iowrite32(v, adev->rio_mem + (mmMM_DATA * 4));
}
if (adev->asic_type >= CHIP_VEGA10 && reg == 1 && adev->last_mm_index == 0x5702C) {
udelay(500);
}
}
/**
@ -318,51 +336,16 @@ static void amdgpu_block_invalid_wreg(struct amdgpu_device *adev,
static int amdgpu_vram_scratch_init(struct amdgpu_device *adev)
{
int r;
if (adev->vram_scratch.robj == NULL) {
r = amdgpu_bo_create(adev, AMDGPU_GPU_PAGE_SIZE,
PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &adev->vram_scratch.robj);
if (r) {
return r;
}
}
r = amdgpu_bo_reserve(adev->vram_scratch.robj, false);
if (unlikely(r != 0))
return r;
r = amdgpu_bo_pin(adev->vram_scratch.robj,
AMDGPU_GEM_DOMAIN_VRAM, &adev->vram_scratch.gpu_addr);
if (r) {
amdgpu_bo_unreserve(adev->vram_scratch.robj);
return r;
}
r = amdgpu_bo_kmap(adev->vram_scratch.robj,
(void **)&adev->vram_scratch.ptr);
if (r)
amdgpu_bo_unpin(adev->vram_scratch.robj);
amdgpu_bo_unreserve(adev->vram_scratch.robj);
return r;
return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
&adev->vram_scratch.robj,
&adev->vram_scratch.gpu_addr,
(void **)&adev->vram_scratch.ptr);
}
static void amdgpu_vram_scratch_fini(struct amdgpu_device *adev)
{
int r;
if (adev->vram_scratch.robj == NULL) {
return;
}
r = amdgpu_bo_reserve(adev->vram_scratch.robj, true);
if (likely(r == 0)) {
amdgpu_bo_kunmap(adev->vram_scratch.robj);
amdgpu_bo_unpin(adev->vram_scratch.robj);
amdgpu_bo_unreserve(adev->vram_scratch.robj);
}
amdgpu_bo_unref(&adev->vram_scratch.robj);
amdgpu_bo_free_kernel(&adev->vram_scratch.robj, NULL, NULL);
}
/**
@ -521,7 +504,8 @@ static int amdgpu_wb_init(struct amdgpu_device *adev)
int r;
if (adev->wb.wb_obj == NULL) {
r = amdgpu_bo_create_kernel(adev, AMDGPU_MAX_WB * sizeof(uint32_t),
/* AMDGPU_MAX_WB * sizeof(uint32_t) * 8 = AMDGPU_MAX_WB 256bit slots */
r = amdgpu_bo_create_kernel(adev, AMDGPU_MAX_WB * sizeof(uint32_t) * 8,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
&adev->wb.wb_obj, &adev->wb.gpu_addr,
(void **)&adev->wb.wb);
@ -552,32 +536,10 @@ static int amdgpu_wb_init(struct amdgpu_device *adev)
int amdgpu_wb_get(struct amdgpu_device *adev, u32 *wb)
{
unsigned long offset = find_first_zero_bit(adev->wb.used, adev->wb.num_wb);
if (offset < adev->wb.num_wb) {
__set_bit(offset, adev->wb.used);
*wb = offset;
return 0;
} else {
return -EINVAL;
}
}
/**
* amdgpu_wb_get_64bit - Allocate a wb entry
*
* @adev: amdgpu_device pointer
* @wb: wb index
*
* Allocate a wb slot for use by the driver (all asics).
* Returns 0 on success or -EINVAL on failure.
*/
int amdgpu_wb_get_64bit(struct amdgpu_device *adev, u32 *wb)
{
unsigned long offset = bitmap_find_next_zero_area_off(adev->wb.used,
adev->wb.num_wb, 0, 2, 7, 0);
if ((offset + 1) < adev->wb.num_wb) {
__set_bit(offset, adev->wb.used);
__set_bit(offset + 1, adev->wb.used);
*wb = offset;
*wb = offset * 8; /* convert to dw offset */
return 0;
} else {
return -EINVAL;
@ -598,22 +560,6 @@ void amdgpu_wb_free(struct amdgpu_device *adev, u32 wb)
__clear_bit(wb, adev->wb.used);
}
/**
* amdgpu_wb_free_64bit - Free a wb entry
*
* @adev: amdgpu_device pointer
* @wb: wb index
*
* Free a wb slot allocated for use by the driver (all asics)
*/
void amdgpu_wb_free_64bit(struct amdgpu_device *adev, u32 wb)
{
if ((wb + 1) < adev->wb.num_wb) {
__clear_bit(wb, adev->wb.used);
__clear_bit(wb + 1, adev->wb.used);
}
}
/**
* amdgpu_vram_location - try to find VRAM location
* @adev: amdgpu device structure holding all necessary informations
@ -665,7 +611,7 @@ void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64
}
/**
* amdgpu_gtt_location - try to find GTT location
* amdgpu_gart_location - try to find GTT location
* @adev: amdgpu device structure holding all necessary informations
* @mc: memory controller structure holding memory informations
*
@ -676,28 +622,28 @@ void amdgpu_vram_location(struct amdgpu_device *adev, struct amdgpu_mc *mc, u64
*
* FIXME: when reducing GTT size align new size on power of 2.
*/
void amdgpu_gtt_location(struct amdgpu_device *adev, struct amdgpu_mc *mc)
void amdgpu_gart_location(struct amdgpu_device *adev, struct amdgpu_mc *mc)
{
u64 size_af, size_bf;
size_af = ((adev->mc.mc_mask - mc->vram_end) + mc->gtt_base_align) & ~mc->gtt_base_align;
size_bf = mc->vram_start & ~mc->gtt_base_align;
size_af = adev->mc.mc_mask - mc->vram_end;
size_bf = mc->vram_start;
if (size_bf > size_af) {
if (mc->gtt_size > size_bf) {
if (mc->gart_size > size_bf) {
dev_warn(adev->dev, "limiting GTT\n");
mc->gtt_size = size_bf;
mc->gart_size = size_bf;
}
mc->gtt_start = 0;
mc->gart_start = 0;
} else {
if (mc->gtt_size > size_af) {
if (mc->gart_size > size_af) {
dev_warn(adev->dev, "limiting GTT\n");
mc->gtt_size = size_af;
mc->gart_size = size_af;
}
mc->gtt_start = (mc->vram_end + 1 + mc->gtt_base_align) & ~mc->gtt_base_align;
mc->gart_start = mc->vram_end + 1;
}
mc->gtt_end = mc->gtt_start + mc->gtt_size - 1;
mc->gart_end = mc->gart_start + mc->gart_size - 1;
dev_info(adev->dev, "GTT: %lluM 0x%016llX - 0x%016llX\n",
mc->gtt_size >> 20, mc->gtt_start, mc->gtt_end);
mc->gart_size >> 20, mc->gart_start, mc->gart_end);
}
/*
@ -720,7 +666,12 @@ bool amdgpu_need_post(struct amdgpu_device *adev)
adev->has_hw_reset = false;
return true;
}
/* then check MEM_SIZE, in case the crtcs are off */
/* bios scratch used on CIK+ */
if (adev->asic_type >= CHIP_BONAIRE)
return amdgpu_atombios_scratch_need_asic_init(adev);
/* check MEM_SIZE for older asics */
reg = amdgpu_asic_get_config_memsize(adev);
if ((reg != 0) && (reg != 0xffffffff))
@ -1031,19 +982,6 @@ static unsigned int amdgpu_vga_set_decode(void *cookie, bool state)
return VGA_RSRC_NORMAL_IO | VGA_RSRC_NORMAL_MEM;
}
/**
* amdgpu_check_pot_argument - check that argument is a power of two
*
* @arg: value to check
*
* Validates that a certain argument is a power of two (all asics).
* Returns true if argument is valid.
*/
static bool amdgpu_check_pot_argument(int arg)
{
return (arg & (arg - 1)) == 0;
}
static void amdgpu_check_block_size(struct amdgpu_device *adev)
{
/* defines number of bits in page table versus page directory,
@ -1077,7 +1015,7 @@ static void amdgpu_check_vm_size(struct amdgpu_device *adev)
if (amdgpu_vm_size == -1)
return;
if (!amdgpu_check_pot_argument(amdgpu_vm_size)) {
if (!is_power_of_2(amdgpu_vm_size)) {
dev_warn(adev->dev, "VM size (%d) must be a power of 2\n",
amdgpu_vm_size);
goto def_value;
@ -1118,19 +1056,24 @@ static void amdgpu_check_arguments(struct amdgpu_device *adev)
dev_warn(adev->dev, "sched jobs (%d) must be at least 4\n",
amdgpu_sched_jobs);
amdgpu_sched_jobs = 4;
} else if (!amdgpu_check_pot_argument(amdgpu_sched_jobs)){
} else if (!is_power_of_2(amdgpu_sched_jobs)){
dev_warn(adev->dev, "sched jobs (%d) must be a power of 2\n",
amdgpu_sched_jobs);
amdgpu_sched_jobs = roundup_pow_of_two(amdgpu_sched_jobs);
}
if (amdgpu_gart_size != -1) {
if (amdgpu_gart_size < 32) {
/* gart size must be greater or equal to 32M */
dev_warn(adev->dev, "gart size (%d) too small\n",
amdgpu_gart_size);
amdgpu_gart_size = 32;
}
if (amdgpu_gtt_size != -1 && amdgpu_gtt_size < 32) {
/* gtt size must be greater or equal to 32M */
if (amdgpu_gart_size < 32) {
dev_warn(adev->dev, "gart size (%d) too small\n",
amdgpu_gart_size);
amdgpu_gart_size = -1;
}
dev_warn(adev->dev, "gtt size (%d) too small\n",
amdgpu_gtt_size);
amdgpu_gtt_size = -1;
}
amdgpu_check_vm_size(adev);
@ -1138,7 +1081,7 @@ static void amdgpu_check_arguments(struct amdgpu_device *adev)
amdgpu_check_block_size(adev);
if (amdgpu_vram_page_split != -1 && (amdgpu_vram_page_split < 16 ||
!amdgpu_check_pot_argument(amdgpu_vram_page_split))) {
!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;
@ -1901,7 +1844,8 @@ static int amdgpu_sriov_reinit_late(struct amdgpu_device *adev)
AMD_IP_BLOCK_TYPE_DCE,
AMD_IP_BLOCK_TYPE_GFX,
AMD_IP_BLOCK_TYPE_SDMA,
AMD_IP_BLOCK_TYPE_VCE,
AMD_IP_BLOCK_TYPE_UVD,
AMD_IP_BLOCK_TYPE_VCE
};
for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
@ -2019,7 +1963,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
adev->flags = flags;
adev->asic_type = flags & AMD_ASIC_MASK;
adev->usec_timeout = AMDGPU_MAX_USEC_TIMEOUT;
adev->mc.gtt_size = 512 * 1024 * 1024;
adev->mc.gart_size = 512 * 1024 * 1024;
adev->accel_working = false;
adev->num_rings = 0;
adev->mman.buffer_funcs = NULL;
@ -2068,6 +2012,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
spin_lock_init(&adev->uvd_ctx_idx_lock);
spin_lock_init(&adev->didt_idx_lock);
spin_lock_init(&adev->gc_cac_idx_lock);
spin_lock_init(&adev->se_cac_idx_lock);
spin_lock_init(&adev->audio_endpt_idx_lock);
spin_lock_init(&adev->mm_stats.lock);
@ -2143,6 +2088,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
r = amdgpu_atombios_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_atombios_init failed\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL, 0, 0);
goto failed;
}
@ -2153,6 +2099,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
if (amdgpu_vpost_needed(adev)) {
if (!adev->bios) {
dev_err(adev->dev, "no vBIOS found\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_NO_VBIOS, 0, 0);
r = -EINVAL;
goto failed;
}
@ -2160,18 +2107,28 @@ int amdgpu_device_init(struct amdgpu_device *adev,
r = amdgpu_atom_asic_init(adev->mode_info.atom_context);
if (r) {
dev_err(adev->dev, "gpu post error!\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_GPU_POST_ERROR, 0, 0);
goto failed;
}
} else {
DRM_INFO("GPU post is not needed\n");
}
if (!adev->is_atom_fw) {
if (adev->is_atom_fw) {
/* Initialize clocks */
r = amdgpu_atomfirmware_get_clock_info(adev);
if (r) {
dev_err(adev->dev, "amdgpu_atomfirmware_get_clock_info failed\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
goto failed;
}
} else {
/* Initialize clocks */
r = amdgpu_atombios_get_clock_info(adev);
if (r) {
dev_err(adev->dev, "amdgpu_atombios_get_clock_info failed\n");
return r;
amdgpu_vf_error_put(AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL, 0, 0);
goto failed;
}
/* init i2c buses */
amdgpu_atombios_i2c_init(adev);
@ -2181,6 +2138,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
r = amdgpu_fence_driver_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_fence_driver_init failed\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_FENCE_INIT_FAIL, 0, 0);
goto failed;
}
@ -2190,6 +2148,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
r = amdgpu_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_init failed\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0);
amdgpu_fini(adev);
goto failed;
}
@ -2209,6 +2168,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
r = amdgpu_ib_pool_init(adev);
if (r) {
dev_err(adev->dev, "IB initialization failed (%d).\n", r);
amdgpu_vf_error_put(AMDGIM_ERROR_VF_IB_INIT_FAIL, 0, r);
goto failed;
}
@ -2253,12 +2213,14 @@ int amdgpu_device_init(struct amdgpu_device *adev,
r = amdgpu_late_init(adev);
if (r) {
dev_err(adev->dev, "amdgpu_late_init failed\n");
amdgpu_vf_error_put(AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL, 0, r);
goto failed;
}
return 0;
failed:
amdgpu_vf_error_trans_all(adev);
if (runtime)
vga_switcheroo_fini_domain_pm_ops(adev->dev);
return r;
@ -2351,6 +2313,8 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
}
drm_modeset_unlock_all(dev);
amdgpu_amdkfd_suspend(adev);
/* unpin the front buffers and cursors */
list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
@ -2392,10 +2356,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
*/
amdgpu_bo_evict_vram(adev);
if (adev->is_atom_fw)
amdgpu_atomfirmware_scratch_regs_save(adev);
else
amdgpu_atombios_scratch_regs_save(adev);
amdgpu_atombios_scratch_regs_save(adev);
pci_save_state(dev->pdev);
if (suspend) {
/* Shut down the device */
@ -2444,10 +2405,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
if (r)
goto unlock;
}
if (adev->is_atom_fw)
amdgpu_atomfirmware_scratch_regs_restore(adev);
else
amdgpu_atombios_scratch_regs_restore(adev);
amdgpu_atombios_scratch_regs_restore(adev);
/* post card */
if (amdgpu_need_post(adev)) {
@ -2490,6 +2448,9 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
}
}
}
r = amdgpu_amdkfd_resume(adev);
if (r)
return r;
/* blat the mode back in */
if (fbcon) {
@ -2860,21 +2821,9 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
r = amdgpu_suspend(adev);
retry:
/* Disable fb access */
if (adev->mode_info.num_crtc) {
struct amdgpu_mode_mc_save save;
amdgpu_display_stop_mc_access(adev, &save);
amdgpu_wait_for_idle(adev, AMD_IP_BLOCK_TYPE_GMC);
}
if (adev->is_atom_fw)
amdgpu_atomfirmware_scratch_regs_save(adev);
else
amdgpu_atombios_scratch_regs_save(adev);
amdgpu_atombios_scratch_regs_save(adev);
r = amdgpu_asic_reset(adev);
if (adev->is_atom_fw)
amdgpu_atomfirmware_scratch_regs_restore(adev);
else
amdgpu_atombios_scratch_regs_restore(adev);
amdgpu_atombios_scratch_regs_restore(adev);
/* post card */
amdgpu_atom_asic_init(adev->mode_info.atom_context);
@ -2952,6 +2901,7 @@ out:
}
} else {
dev_err(adev->dev, "asic resume failed (%d).\n", r);
amdgpu_vf_error_put(AMDGIM_ERROR_VF_ASIC_RESUME_FAIL, 0, r);
for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
if (adev->rings[i] && adev->rings[i]->sched.thread) {
kthread_unpark(adev->rings[i]->sched.thread);
@ -2962,12 +2912,16 @@ out:
drm_helper_resume_force_mode(adev->ddev);
ttm_bo_unlock_delayed_workqueue(&adev->mman.bdev, resched);
if (r)
if (r) {
/* bad news, how to tell it to userspace ? */
dev_info(adev->dev, "GPU reset failed\n");
else
amdgpu_vf_error_put(AMDGIM_ERROR_VF_GPU_RESET_FAIL, 0, r);
}
else {
dev_info(adev->dev, "GPU reset successed!\n");
}
amdgpu_vf_error_trans_all(adev);
return r;
}

6
drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
View file

@ -482,7 +482,7 @@ static void amdgpu_user_framebuffer_destroy(struct drm_framebuffer *fb)
{
struct amdgpu_framebuffer *amdgpu_fb = to_amdgpu_framebuffer(fb);
drm_gem_object_unreference_unlocked(amdgpu_fb->obj);
drm_gem_object_put_unlocked(amdgpu_fb->obj);
drm_framebuffer_cleanup(fb);
kfree(amdgpu_fb);
}
@ -542,14 +542,14 @@ amdgpu_user_framebuffer_create(struct drm_device *dev,
amdgpu_fb = kzalloc(sizeof(*amdgpu_fb), GFP_KERNEL);
if (amdgpu_fb == NULL) {
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ERR_PTR(-ENOMEM);
}
ret = amdgpu_framebuffer_init(dev, amdgpu_fb, mode_cmd, obj);
if (ret) {
kfree(amdgpu_fb);
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ERR_PTR(ret);
}

22
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
View file

@ -74,7 +74,9 @@
#define KMS_DRIVER_PATCHLEVEL 0
int amdgpu_vram_limit = 0;
int amdgpu_gart_size = -1; /* auto */
int amdgpu_vis_vram_limit = 0;
unsigned amdgpu_gart_size = 256;
int amdgpu_gtt_size = -1; /* auto */
int amdgpu_moverate = -1; /* auto */
int amdgpu_benchmarking = 0;
int amdgpu_testing = 0;
@ -106,6 +108,7 @@ unsigned amdgpu_pcie_gen_cap = 0;
unsigned amdgpu_pcie_lane_cap = 0;
unsigned amdgpu_cg_mask = 0xffffffff;
unsigned amdgpu_pg_mask = 0xffffffff;
unsigned amdgpu_sdma_phase_quantum = 32;
char *amdgpu_disable_cu = NULL;
char *amdgpu_virtual_display = NULL;
unsigned amdgpu_pp_feature_mask = 0xffffffff;
@ -120,8 +123,14 @@ int amdgpu_lbpw = -1;
MODULE_PARM_DESC(vramlimit, "Restrict VRAM for testing, in megabytes");
module_param_named(vramlimit, amdgpu_vram_limit, int, 0600);
MODULE_PARM_DESC(gartsize, "Size of PCIE/IGP gart to setup in megabytes (32, 64, etc., -1 = auto)");
module_param_named(gartsize, amdgpu_gart_size, int, 0600);
MODULE_PARM_DESC(vis_vramlimit, "Restrict visible VRAM for testing, in megabytes");
module_param_named(vis_vramlimit, amdgpu_vis_vram_limit, int, 0444);
MODULE_PARM_DESC(gartsize, "Size of PCIE/IGP gart to setup in megabytes (32, 64, etc.)");
module_param_named(gartsize, amdgpu_gart_size, uint, 0600);
MODULE_PARM_DESC(gttsize, "Size of the GTT domain in megabytes (-1 = auto)");
module_param_named(gttsize, amdgpu_gtt_size, int, 0600);
MODULE_PARM_DESC(moverate, "Maximum buffer migration rate in MB/s. (32, 64, etc., -1=auto, 0=1=disabled)");
module_param_named(moverate, amdgpu_moverate, int, 0600);
@ -186,7 +195,7 @@ 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);
MODULE_PARM_DESC(vram_page_split, "Number of pages after we split VRAM allocations (default 1024, -1 = disable)");
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);
MODULE_PARM_DESC(exp_hw_support, "experimental hw support (1 = enable, 0 = disable (default))");
@ -199,7 +208,7 @@ MODULE_PARM_DESC(sched_hw_submission, "the max number of HW submissions (default
module_param_named(sched_hw_submission, amdgpu_sched_hw_submission, int, 0444);
MODULE_PARM_DESC(ppfeaturemask, "all power features enabled (default))");
module_param_named(ppfeaturemask, amdgpu_pp_feature_mask, int, 0444);
module_param_named(ppfeaturemask, amdgpu_pp_feature_mask, uint, 0444);
MODULE_PARM_DESC(no_evict, "Support pinning request from user space (1 = enable, 0 = disable (default))");
module_param_named(no_evict, amdgpu_no_evict, int, 0444);
@ -219,6 +228,9 @@ module_param_named(cg_mask, amdgpu_cg_mask, uint, 0444);
MODULE_PARM_DESC(pg_mask, "Powergating flags mask (0 = disable power gating)");
module_param_named(pg_mask, amdgpu_pg_mask, uint, 0444);
MODULE_PARM_DESC(sdma_phase_quantum, "SDMA context switch phase quantum (x 1K GPU clock cycles, 0 = no change (default 32))");
module_param_named(sdma_phase_quantum, amdgpu_sdma_phase_quantum, uint, 0444);
MODULE_PARM_DESC(disable_cu, "Disable CUs (se.sh.cu,...)");
module_param_named(disable_cu, amdgpu_disable_cu, charp, 0444);

6
drivers/gpu/drm/amd/amdgpu/amdgpu_fb.c
View file

@ -118,7 +118,7 @@ static void amdgpufb_destroy_pinned_object(struct drm_gem_object *gobj)
amdgpu_bo_unpin(abo);
amdgpu_bo_unreserve(abo);
}
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
}
static int amdgpufb_create_pinned_object(struct amdgpu_fbdev *rfbdev,
@ -250,7 +250,7 @@ static int amdgpufb_create(struct drm_fb_helper *helper,
tmp = amdgpu_bo_gpu_offset(abo) - adev->mc.vram_start;
info->fix.smem_start = adev->mc.aper_base + tmp;
info->fix.smem_len = amdgpu_bo_size(abo);
info->screen_base = abo->kptr;
info->screen_base = amdgpu_bo_kptr(abo);
info->screen_size = amdgpu_bo_size(abo);
drm_fb_helper_fill_var(info, &rfbdev->helper, sizes->fb_width, sizes->fb_height);
@ -280,7 +280,7 @@ out:
}
if (fb && ret) {
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
drm_framebuffer_unregister_private(fb);
drm_framebuffer_cleanup(fb);
kfree(fb);

81
drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c
View file

@ -55,6 +55,19 @@
/*
* Common GART table functions.
*/
/**
* amdgpu_gart_set_defaults - set the default gart_size
*
* @adev: amdgpu_device pointer
*
* Set the default gart_size based on parameters and available VRAM.
*/
void amdgpu_gart_set_defaults(struct amdgpu_device *adev)
{
adev->mc.gart_size = (uint64_t)amdgpu_gart_size << 20;
}
/**
* amdgpu_gart_table_ram_alloc - allocate system ram for gart page table
*
@ -131,7 +144,7 @@ int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev)
PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &adev->gart.robj);
NULL, NULL, 0, &adev->gart.robj);
if (r) {
return r;
}
@ -262,6 +275,41 @@ int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
return 0;
}
/**
* amdgpu_gart_map - map dma_addresses into GART entries
*
* @adev: amdgpu_device pointer
* @offset: offset into the GPU's gart aperture
* @pages: number of pages to bind
* @dma_addr: DMA addresses of pages
*
* Map the dma_addresses into GART entries (all asics).
* Returns 0 for success, -EINVAL for failure.
*/
int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
int pages, dma_addr_t *dma_addr, uint64_t flags,
void *dst)
{
uint64_t page_base;
unsigned i, j, t;
if (!adev->gart.ready) {
WARN(1, "trying to bind memory to uninitialized GART !\n");
return -EINVAL;
}
t = offset / AMDGPU_GPU_PAGE_SIZE;
for (i = 0; i < pages; i++) {
page_base = dma_addr[i];
for (j = 0; j < (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE); j++, t++) {
amdgpu_gart_set_pte_pde(adev, dst, t, page_base, flags);
page_base += AMDGPU_GPU_PAGE_SIZE;
}
}
return 0;
}
/**
* amdgpu_gart_bind - bind pages into the gart page table
*
@ -279,31 +327,30 @@ int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
int pages, struct page **pagelist, dma_addr_t *dma_addr,
uint64_t flags)
{
unsigned t;
unsigned p;
uint64_t page_base;
int i, j;
#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
unsigned i,t,p;
#endif
int r;
if (!adev->gart.ready) {
WARN(1, "trying to bind memory to uninitialized GART !\n");
return -EINVAL;
}
#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
t = offset / AMDGPU_GPU_PAGE_SIZE;
p = t / (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE);
for (i = 0; i < pages; i++, p++) {
#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
for (i = 0; i < pages; i++, p++)
adev->gart.pages[p] = pagelist[i];
#endif
if (adev->gart.ptr) {
page_base = dma_addr[i];
for (j = 0; j < (PAGE_SIZE / AMDGPU_GPU_PAGE_SIZE); j++, t++) {
amdgpu_gart_set_pte_pde(adev, adev->gart.ptr, t, page_base, flags);
page_base += AMDGPU_GPU_PAGE_SIZE;
}
}
if (adev->gart.ptr) {
r = amdgpu_gart_map(adev, offset, pages, dma_addr, flags,
adev->gart.ptr);
if (r)
return r;
}
mb();
amdgpu_gart_flush_gpu_tlb(adev, 0);
return 0;
@ -333,8 +380,8 @@ int amdgpu_gart_init(struct amdgpu_device *adev)
if (r)
return r;
/* Compute table size */
adev->gart.num_cpu_pages = adev->mc.gtt_size / PAGE_SIZE;
adev->gart.num_gpu_pages = adev->mc.gtt_size / AMDGPU_GPU_PAGE_SIZE;
adev->gart.num_cpu_pages = adev->mc.gart_size / PAGE_SIZE;
adev->gart.num_gpu_pages = adev->mc.gart_size / AMDGPU_GPU_PAGE_SIZE;
DRM_INFO("GART: num cpu pages %u, num gpu pages %u\n",
adev->gart.num_cpu_pages, adev->gart.num_gpu_pages);

77
drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h Normal file
View file

@ -0,0 +1,77 @@
/*
* Copyright 2017 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_GART_H__
#define __AMDGPU_GART_H__
#include <linux/types.h>
/*
* GART structures, functions & helpers
*/
struct amdgpu_device;
struct amdgpu_bo;
struct amdgpu_gart_funcs;
#define AMDGPU_GPU_PAGE_SIZE 4096
#define AMDGPU_GPU_PAGE_MASK (AMDGPU_GPU_PAGE_SIZE - 1)
#define AMDGPU_GPU_PAGE_SHIFT 12
#define AMDGPU_GPU_PAGE_ALIGN(a) (((a) + AMDGPU_GPU_PAGE_MASK) & ~AMDGPU_GPU_PAGE_MASK)
struct amdgpu_gart {
dma_addr_t table_addr;
struct amdgpu_bo *robj;
void *ptr;
unsigned num_gpu_pages;
unsigned num_cpu_pages;
unsigned table_size;
#ifdef CONFIG_DRM_AMDGPU_GART_DEBUGFS
struct page **pages;
#endif
bool ready;
/* Asic default pte flags */
uint64_t gart_pte_flags;
const struct amdgpu_gart_funcs *gart_funcs;
};
void amdgpu_gart_set_defaults(struct amdgpu_device *adev);
int amdgpu_gart_table_ram_alloc(struct amdgpu_device *adev);
void amdgpu_gart_table_ram_free(struct amdgpu_device *adev);
int amdgpu_gart_table_vram_alloc(struct amdgpu_device *adev);
void amdgpu_gart_table_vram_free(struct amdgpu_device *adev);
int amdgpu_gart_table_vram_pin(struct amdgpu_device *adev);
void amdgpu_gart_table_vram_unpin(struct amdgpu_device *adev);
int amdgpu_gart_init(struct amdgpu_device *adev);
void amdgpu_gart_fini(struct amdgpu_device *adev);
int amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset,
int pages);
int amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset,
int pages, dma_addr_t *dma_addr, uint64_t flags,
void *dst);
int amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset,
int pages, struct page **pagelist,
dma_addr_t *dma_addr, uint64_t flags);
#endif

56
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
View file

@ -49,7 +49,6 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
struct drm_gem_object **obj)
{
struct amdgpu_bo *robj;
unsigned long max_size;
int r;
*obj = NULL;
@ -58,20 +57,9 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
alignment = PAGE_SIZE;
}
if (!(initial_domain & (AMDGPU_GEM_DOMAIN_GDS | AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA))) {
/* Maximum bo size is the unpinned gtt size since we use the gtt to
* handle vram to system pool migrations.
*/
max_size = adev->mc.gtt_size - adev->gart_pin_size;
if (size > max_size) {
DRM_DEBUG("Allocation size %ldMb bigger than %ldMb limit\n",
size >> 20, max_size >> 20);
return -ENOMEM;
}
}
retry:
r = amdgpu_bo_create(adev, size, alignment, kernel, initial_domain,
flags, NULL, NULL, &robj);
flags, NULL, NULL, 0, &robj);
if (r) {
if (r != -ERESTARTSYS) {
if (initial_domain == AMDGPU_GEM_DOMAIN_VRAM) {
@ -103,7 +91,7 @@ void amdgpu_gem_force_release(struct amdgpu_device *adev)
spin_lock(&file->table_lock);
idr_for_each_entry(&file->object_idr, gobj, handle) {
WARN_ONCE(1, "And also active allocations!\n");
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
}
idr_destroy(&file->object_idr);
spin_unlock(&file->table_lock);
@ -275,7 +263,7 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
r = drm_gem_handle_create(filp, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
if (r)
return r;
@ -318,7 +306,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
return r;
bo = gem_to_amdgpu_bo(gobj);
bo->prefered_domains = AMDGPU_GEM_DOMAIN_GTT;
bo->preferred_domains = AMDGPU_GEM_DOMAIN_GTT;
bo->allowed_domains = AMDGPU_GEM_DOMAIN_GTT;
r = amdgpu_ttm_tt_set_userptr(bo->tbo.ttm, args->addr, args->flags);
if (r)
@ -353,7 +341,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
r = drm_gem_handle_create(filp, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
if (r)
return r;
@ -367,7 +355,7 @@ unlock_mmap_sem:
up_read(&current->mm->mmap_sem);
release_object:
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return r;
}
@ -386,11 +374,11 @@ int amdgpu_mode_dumb_mmap(struct drm_file *filp,
robj = gem_to_amdgpu_bo(gobj);
if (amdgpu_ttm_tt_get_usermm(robj->tbo.ttm) ||
(robj->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)) {
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return -EPERM;
}
*offset_p = amdgpu_bo_mmap_offset(robj);
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return 0;
}
@ -460,7 +448,7 @@ int amdgpu_gem_wait_idle_ioctl(struct drm_device *dev, void *data,
} else
r = ret;
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return r;
}
@ -503,7 +491,7 @@ int amdgpu_gem_metadata_ioctl(struct drm_device *dev, void *data,
unreserve:
amdgpu_bo_unreserve(robj);
out:
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return r;
}
@ -676,7 +664,7 @@ error_backoff:
ttm_eu_backoff_reservation(&ticket, &list);
error_unref:
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return r;
}
@ -701,11 +689,11 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
switch (args->op) {
case AMDGPU_GEM_OP_GET_GEM_CREATE_INFO: {
struct drm_amdgpu_gem_create_in info;
void __user *out = (void __user *)(uintptr_t)args->value;
void __user *out = u64_to_user_ptr(args->value);
info.bo_size = robj->gem_base.size;
info.alignment = robj->tbo.mem.page_alignment << PAGE_SHIFT;
info.domains = robj->prefered_domains;
info.domains = robj->preferred_domains;
info.domain_flags = robj->flags;
amdgpu_bo_unreserve(robj);
if (copy_to_user(out, &info, sizeof(info)))
@ -723,10 +711,10 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
amdgpu_bo_unreserve(robj);
break;
}
robj->prefered_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
robj->preferred_domains = args->value & (AMDGPU_GEM_DOMAIN_VRAM |
AMDGPU_GEM_DOMAIN_GTT |
AMDGPU_GEM_DOMAIN_CPU);
robj->allowed_domains = robj->prefered_domains;
robj->allowed_domains = robj->preferred_domains;
if (robj->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
robj->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
@ -738,7 +726,7 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data,
}
out:
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
return r;
}
@ -766,7 +754,7 @@ int amdgpu_mode_dumb_create(struct drm_file *file_priv,
r = drm_gem_handle_create(file_priv, gobj, &handle);
/* drop reference from allocate - handle holds it now */
drm_gem_object_unreference_unlocked(gobj);
drm_gem_object_put_unlocked(gobj);
if (r) {
return r;
}
@ -784,6 +772,7 @@ static int amdgpu_debugfs_gem_bo_info(int id, void *ptr, void *data)
unsigned domain;
const char *placement;
unsigned pin_count;
uint64_t offset;
domain = amdgpu_mem_type_to_domain(bo->tbo.mem.mem_type);
switch (domain) {
@ -798,9 +787,12 @@ static int amdgpu_debugfs_gem_bo_info(int id, void *ptr, void *data)
placement = " CPU";
break;
}
seq_printf(m, "\t0x%08x: %12ld byte %s @ 0x%010Lx",
id, amdgpu_bo_size(bo), placement,
amdgpu_bo_gpu_offset(bo));
seq_printf(m, "\t0x%08x: %12ld byte %s",
id, amdgpu_bo_size(bo), placement);
offset = ACCESS_ONCE(bo->tbo.mem.start);
if (offset != AMDGPU_BO_INVALID_OFFSET)
seq_printf(m, " @ 0x%010Lx", offset);
pin_count = ACCESS_ONCE(bo->pin_count);
if (pin_count)

3
drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
View file

@ -125,7 +125,8 @@ void amdgpu_gfx_compute_queue_acquire(struct amdgpu_device *adev)
if (mec >= adev->gfx.mec.num_mec)
break;
if (adev->gfx.mec.num_mec > 1) {
/* FIXME: spreading the queues across pipes causes perf regressions */
if (0) {
/* policy: amdgpu owns the first two queues of the first MEC */
if (mec == 0 && queue < 2)
set_bit(i, adev->gfx.mec.queue_bitmap);

25
drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c
View file

@ -42,13 +42,17 @@ struct amdgpu_gtt_mgr {
static int amdgpu_gtt_mgr_init(struct ttm_mem_type_manager *man,
unsigned long p_size)
{
struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
struct amdgpu_gtt_mgr *mgr;
uint64_t start, size;
mgr = kzalloc(sizeof(*mgr), GFP_KERNEL);
if (!mgr)
return -ENOMEM;
drm_mm_init(&mgr->mm, 0, p_size);
start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS;
size = (adev->mc.gart_size >> PAGE_SHIFT) - start;
drm_mm_init(&mgr->mm, start, size);
spin_lock_init(&mgr->lock);
mgr->available = p_size;
man->priv = mgr;
@ -80,6 +84,20 @@ static int amdgpu_gtt_mgr_fini(struct ttm_mem_type_manager *man)
return 0;
}
/**
* amdgpu_gtt_mgr_is_allocated - Check if mem has address space
*
* @mem: the mem object to check
*
* Check if a mem object has already address space allocated.
*/
bool amdgpu_gtt_mgr_is_allocated(struct ttm_mem_reg *mem)
{
struct drm_mm_node *node = mem->mm_node;
return (node->start != AMDGPU_BO_INVALID_OFFSET);
}
/**
* amdgpu_gtt_mgr_alloc - allocate new ranges
*
@ -95,13 +113,14 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
const struct ttm_place *place,
struct ttm_mem_reg *mem)
{
struct amdgpu_device *adev = amdgpu_ttm_adev(man->bdev);
struct amdgpu_gtt_mgr *mgr = man->priv;
struct drm_mm_node *node = mem->mm_node;
enum drm_mm_insert_mode mode;
unsigned long fpfn, lpfn;
int r;
if (node->start != AMDGPU_BO_INVALID_OFFSET)
if (amdgpu_gtt_mgr_is_allocated(mem))
return 0;
if (place)
@ -112,7 +131,7 @@ int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
if (place && place->lpfn)
lpfn = place->lpfn;
else
lpfn = man->size;
lpfn = adev->gart.num_cpu_pages;
mode = DRM_MM_INSERT_BEST;
if (place && place->flags & TTM_PL_FLAG_TOPDOWN)

7
drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
View file

@ -130,6 +130,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
unsigned i;
int r = 0;
bool need_pipe_sync = false;
if (num_ibs == 0)
return -EINVAL;
@ -165,15 +166,15 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
if (ring->funcs->emit_pipeline_sync && job &&
((tmp = amdgpu_sync_get_fence(&job->sched_sync)) ||
amdgpu_vm_need_pipeline_sync(ring, job))) {
amdgpu_ring_emit_pipeline_sync(ring);
need_pipe_sync = true;
dma_fence_put(tmp);
}
if (ring->funcs->insert_start)
ring->funcs->insert_start(ring);
if (vm) {
r = amdgpu_vm_flush(ring, job);
if (job) {
r = amdgpu_vm_flush(ring, job, need_pipe_sync);
if (r) {
amdgpu_ring_undo(ring);
return r;

4
drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
View file

@ -220,6 +220,10 @@ int amdgpu_irq_init(struct amdgpu_device *adev)
int r = 0;
spin_lock_init(&adev->irq.lock);
/* Disable vblank irqs aggressively for power-saving */
adev->ddev->vblank_disable_immediate = true;
r = drm_vblank_init(adev->ddev, adev->mode_info.num_crtc);
if (r) {
return r;

2
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
View file

@ -81,6 +81,8 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
r = amdgpu_ib_get(adev, NULL, size, &(*job)->ibs[0]);
if (r)
kfree(*job);
else
(*job)->vm_pd_addr = adev->gart.table_addr;
return r;
}

20
drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
View file

@ -158,7 +158,6 @@ int amdgpu_driver_load_kms(struct drm_device *dev, unsigned long flags)
"Error during ACPI methods call\n");
}
amdgpu_amdkfd_load_interface(adev);
amdgpu_amdkfd_device_probe(adev);
amdgpu_amdkfd_device_init(adev);
@ -485,7 +484,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
vram_gtt.vram_size -= adev->vram_pin_size;
vram_gtt.vram_cpu_accessible_size = adev->mc.visible_vram_size;
vram_gtt.vram_cpu_accessible_size -= (adev->vram_pin_size - adev->invisible_pin_size);
vram_gtt.gtt_size = adev->mc.gtt_size;
vram_gtt.gtt_size = adev->mman.bdev.man[TTM_PL_TT].size;
vram_gtt.gtt_size *= PAGE_SIZE;
vram_gtt.gtt_size -= adev->gart_pin_size;
return copy_to_user(out, &vram_gtt,
min((size_t)size, sizeof(vram_gtt))) ? -EFAULT : 0;
@ -510,9 +510,10 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
mem.cpu_accessible_vram.max_allocation =
mem.cpu_accessible_vram.usable_heap_size * 3 / 4;
mem.gtt.total_heap_size = adev->mc.gtt_size;
mem.gtt.usable_heap_size =
adev->mc.gtt_size - adev->gart_pin_size;
mem.gtt.total_heap_size = adev->mman.bdev.man[TTM_PL_TT].size;
mem.gtt.total_heap_size *= PAGE_SIZE;
mem.gtt.usable_heap_size = mem.gtt.total_heap_size
- adev->gart_pin_size;
mem.gtt.heap_usage = atomic64_read(&adev->gtt_usage);
mem.gtt.max_allocation = mem.gtt.usable_heap_size * 3 / 4;
@ -571,8 +572,8 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
dev_info.max_engine_clock = amdgpu_dpm_get_sclk(adev, false) * 10;
dev_info.max_memory_clock = amdgpu_dpm_get_mclk(adev, false) * 10;
} else {
dev_info.max_engine_clock = adev->pm.default_sclk * 10;
dev_info.max_memory_clock = adev->pm.default_mclk * 10;
dev_info.max_engine_clock = adev->clock.default_sclk * 10;
dev_info.max_memory_clock = adev->clock.default_mclk * 10;
}
dev_info.enabled_rb_pipes_mask = adev->gfx.config.backend_enable_mask;
dev_info.num_rb_pipes = adev->gfx.config.max_backends_per_se *
@ -587,8 +588,9 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
dev_info.virtual_address_offset = AMDGPU_VA_RESERVED_SIZE;
dev_info.virtual_address_max = (uint64_t)adev->vm_manager.max_pfn * AMDGPU_GPU_PAGE_SIZE;
dev_info.virtual_address_alignment = max((int)PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE);
dev_info.pte_fragment_size = (1 << AMDGPU_LOG2_PAGES_PER_FRAG) *
AMDGPU_GPU_PAGE_SIZE;
dev_info.pte_fragment_size =
(1 << AMDGPU_LOG2_PAGES_PER_FRAG(adev)) *
AMDGPU_GPU_PAGE_SIZE;
dev_info.gart_page_size = AMDGPU_GPU_PAGE_SIZE;
dev_info.cu_active_number = adev->gfx.cu_info.number;

2
drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c
View file

@ -359,7 +359,7 @@ void amdgpu_mn_unregister(struct amdgpu_bo *bo)
head = bo->mn_list.next;
bo->mn = NULL;
list_del(&bo->mn_list);
list_del_init(&bo->mn_list);
if (list_empty(head)) {
struct amdgpu_mn_node *node;

12
drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h
View file

@ -257,15 +257,7 @@ struct amdgpu_audio {
int num_pins;
};
struct amdgpu_mode_mc_save {
u32 vga_render_control;
u32 vga_hdp_control;
bool crtc_enabled[AMDGPU_MAX_CRTCS];
};
struct amdgpu_display_funcs {
/* vga render */
void (*set_vga_render_state)(struct amdgpu_device *adev, bool render);
/* display watermarks */
void (*bandwidth_update)(struct amdgpu_device *adev);
/* get frame count */
@ -300,10 +292,6 @@ struct amdgpu_display_funcs {
uint16_t connector_object_id,
struct amdgpu_hpd *hpd,
struct amdgpu_router *router);
void (*stop_mc_access)(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save);
void (*resume_mc_access)(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save);
};
struct amdgpu_mode_info {

182
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
View file

@ -93,6 +93,7 @@ static void amdgpu_ttm_bo_destroy(struct ttm_buffer_object *tbo)
bo = container_of(tbo, struct amdgpu_bo, tbo);
amdgpu_bo_kunmap(bo);
amdgpu_update_memory_usage(adev, &bo->tbo.mem, NULL);
drm_gem_object_release(&bo->gem_base);
@ -219,7 +220,7 @@ static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
}
/**
* amdgpu_bo_create_kernel - create BO for kernel use
* amdgpu_bo_create_reserved - create reserved BO for kernel use
*
* @adev: amdgpu device object
* @size: size for the new BO
@ -229,24 +230,30 @@ static void amdgpu_fill_placement_to_bo(struct amdgpu_bo *bo,
* @gpu_addr: GPU addr of the pinned BO
* @cpu_addr: optional CPU address mapping
*
* Allocates and pins a BO for kernel internal use.
* Allocates and pins a BO for kernel internal use, and returns it still
* reserved.
*
* Returns 0 on success, negative error code otherwise.
*/
int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
unsigned long size, int align,
u32 domain, struct amdgpu_bo **bo_ptr,
u64 *gpu_addr, void **cpu_addr)
int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
unsigned long size, int align,
u32 domain, struct amdgpu_bo **bo_ptr,
u64 *gpu_addr, void **cpu_addr)
{
bool free = false;
int r;
r = amdgpu_bo_create(adev, size, align, true, domain,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, bo_ptr);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate kernel bo\n", r);
return r;
if (!*bo_ptr) {
r = amdgpu_bo_create(adev, size, align, true, domain,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, 0, bo_ptr);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate kernel bo\n",
r);
return r;
}
free = true;
}
r = amdgpu_bo_reserve(*bo_ptr, false);
@ -269,19 +276,51 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
}
}
amdgpu_bo_unreserve(*bo_ptr);
return 0;
error_unreserve:
amdgpu_bo_unreserve(*bo_ptr);
error_free:
amdgpu_bo_unref(bo_ptr);
if (free)
amdgpu_bo_unref(bo_ptr);
return r;
}
/**
* amdgpu_bo_create_kernel - create BO for kernel use
*
* @adev: amdgpu device object
* @size: size for the new BO
* @align: alignment for the new BO
* @domain: where to place it
* @bo_ptr: resulting BO
* @gpu_addr: GPU addr of the pinned BO
* @cpu_addr: optional CPU address mapping
*
* Allocates and pins a BO for kernel internal use.
*
* Returns 0 on success, negative error code otherwise.
*/
int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
unsigned long size, int align,
u32 domain, struct amdgpu_bo **bo_ptr,
u64 *gpu_addr, void **cpu_addr)
{
int r;
r = amdgpu_bo_create_reserved(adev, size, align, domain, bo_ptr,
gpu_addr, cpu_addr);
if (r)
return r;
amdgpu_bo_unreserve(*bo_ptr);
return 0;
}
/**
* amdgpu_bo_free_kernel - free BO for kernel use
*
@ -317,12 +356,13 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
struct sg_table *sg,
struct ttm_placement *placement,
struct reservation_object *resv,
uint64_t init_value,
struct amdgpu_bo **bo_ptr)
{
struct amdgpu_bo *bo;
enum ttm_bo_type type;
unsigned long page_align;
u64 initial_bytes_moved;
u64 initial_bytes_moved, bytes_moved;
size_t acc_size;
int r;
@ -351,13 +391,13 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
}
INIT_LIST_HEAD(&bo->shadow_list);
INIT_LIST_HEAD(&bo->va);
bo->prefered_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM |
bo->preferred_domains = domain & (AMDGPU_GEM_DOMAIN_VRAM |
AMDGPU_GEM_DOMAIN_GTT |
AMDGPU_GEM_DOMAIN_CPU |
AMDGPU_GEM_DOMAIN_GDS |
AMDGPU_GEM_DOMAIN_GWS |
AMDGPU_GEM_DOMAIN_OA);
bo->allowed_domains = bo->prefered_domains;
bo->allowed_domains = bo->preferred_domains;
if (!kernel && bo->allowed_domains == AMDGPU_GEM_DOMAIN_VRAM)
bo->allowed_domains |= AMDGPU_GEM_DOMAIN_GTT;
@ -398,8 +438,14 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, type,
&bo->placement, page_align, !kernel, NULL,
acc_size, sg, resv, &amdgpu_ttm_bo_destroy);
amdgpu_cs_report_moved_bytes(adev,
atomic64_read(&adev->num_bytes_moved) - initial_bytes_moved);
bytes_moved = atomic64_read(&adev->num_bytes_moved) -
initial_bytes_moved;
if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
bo->tbo.mem.mem_type == TTM_PL_VRAM &&
bo->tbo.mem.start < adev->mc.visible_vram_size >> PAGE_SHIFT)
amdgpu_cs_report_moved_bytes(adev, bytes_moved, bytes_moved);
else
amdgpu_cs_report_moved_bytes(adev, bytes_moved, 0);
if (unlikely(r != 0))
return r;
@ -411,7 +457,7 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
bo->tbo.mem.placement & TTM_PL_FLAG_VRAM) {
struct dma_fence *fence;
r = amdgpu_fill_buffer(bo, 0, bo->tbo.resv, &fence);
r = amdgpu_fill_buffer(bo, init_value, bo->tbo.resv, &fence);
if (unlikely(r))
goto fail_unreserve;
@ -426,6 +472,10 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
trace_amdgpu_bo_create(bo);
/* Treat CPU_ACCESS_REQUIRED only as a hint if given by UMD */
if (type == ttm_bo_type_device)
bo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
return 0;
fail_unreserve:
@ -459,6 +509,7 @@ static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
AMDGPU_GEM_CREATE_CPU_GTT_USWC,
NULL, &placement,
bo->tbo.resv,
0,
&bo->shadow);
if (!r) {
bo->shadow->parent = amdgpu_bo_ref(bo);
@ -470,11 +521,15 @@ static int amdgpu_bo_create_shadow(struct amdgpu_device *adev,
return r;
}
/* init_value will only take effect when flags contains
* AMDGPU_GEM_CREATE_VRAM_CLEARED.
*/
int amdgpu_bo_create(struct amdgpu_device *adev,
unsigned long size, int byte_align,
bool kernel, u32 domain, u64 flags,
struct sg_table *sg,
struct reservation_object *resv,
uint64_t init_value,
struct amdgpu_bo **bo_ptr)
{
struct ttm_placement placement = {0};
@ -489,7 +544,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev,
r = amdgpu_bo_create_restricted(adev, size, byte_align, kernel,
domain, flags, sg, &placement,
resv, bo_ptr);
resv, init_value, bo_ptr);
if (r)
return r;
@ -535,7 +590,7 @@ int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
r = amdgpu_copy_buffer(ring, bo_addr, shadow_addr,
amdgpu_bo_size(bo), resv, fence,
direct);
direct, false);
if (!r)
amdgpu_bo_fence(bo, *fence, true);
@ -551,7 +606,7 @@ int amdgpu_bo_validate(struct amdgpu_bo *bo)
if (bo->pin_count)
return 0;
domain = bo->prefered_domains;
domain = bo->preferred_domains;
retry:
amdgpu_ttm_placement_from_domain(bo, domain);
@ -588,7 +643,7 @@ int amdgpu_bo_restore_from_shadow(struct amdgpu_device *adev,
r = amdgpu_copy_buffer(ring, shadow_addr, bo_addr,
amdgpu_bo_size(bo), resv, fence,
direct);
direct, false);
if (!r)
amdgpu_bo_fence(bo, *fence, true);
@ -598,16 +653,16 @@ err:
int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
{
bool is_iomem;
void *kptr;
long r;
if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
return -EPERM;
if (bo->kptr) {
if (ptr) {
*ptr = bo->kptr;
}
kptr = amdgpu_bo_kptr(bo);
if (kptr) {
if (ptr)
*ptr = kptr;
return 0;
}
@ -620,19 +675,23 @@ int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr)
if (r)
return r;
bo->kptr = ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
if (ptr)
*ptr = bo->kptr;
*ptr = amdgpu_bo_kptr(bo);
return 0;
}
void *amdgpu_bo_kptr(struct amdgpu_bo *bo)
{
bool is_iomem;
return ttm_kmap_obj_virtual(&bo->kmap, &is_iomem);
}
void amdgpu_bo_kunmap(struct amdgpu_bo *bo)
{
if (bo->kptr == NULL)
return;
bo->kptr = NULL;
ttm_bo_kunmap(&bo->kmap);
if (bo->kmap.bo)
ttm_bo_kunmap(&bo->kmap);
}
struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo)
@ -724,15 +783,16 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
dev_err(adev->dev, "%p pin failed\n", bo);
goto error;
}
r = amdgpu_ttm_bind(&bo->tbo, &bo->tbo.mem);
if (unlikely(r)) {
dev_err(adev->dev, "%p bind failed\n", bo);
goto error;
}
bo->pin_count = 1;
if (gpu_addr != NULL)
if (gpu_addr != NULL) {
r = amdgpu_ttm_bind(&bo->tbo, &bo->tbo.mem);
if (unlikely(r)) {
dev_err(adev->dev, "%p bind failed\n", bo);
goto error;
}
*gpu_addr = amdgpu_bo_gpu_offset(bo);
}
if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
adev->vram_pin_size += amdgpu_bo_size(bo);
if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS)
@ -921,6 +981,8 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
abo = container_of(bo, struct amdgpu_bo, tbo);
amdgpu_vm_bo_invalidate(adev, abo);
amdgpu_bo_kunmap(abo);
/* remember the eviction */
if (evict)
atomic64_inc(&adev->num_evictions);
@ -939,19 +1001,22 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
{
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
struct amdgpu_bo *abo;
unsigned long offset, size, lpfn;
int i, r;
unsigned long offset, size;
int r;
if (!amdgpu_ttm_bo_is_amdgpu_bo(bo))
return 0;
abo = container_of(bo, struct amdgpu_bo, tbo);
/* Remember that this BO was accessed by the CPU */
abo->flags |= AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
if (bo->mem.mem_type != TTM_PL_VRAM)
return 0;
size = bo->mem.num_pages << PAGE_SHIFT;
offset = bo->mem.start << PAGE_SHIFT;
/* TODO: figure out how to map scattered VRAM to the CPU */
if ((offset + size) <= adev->mc.visible_vram_size)
return 0;
@ -961,26 +1026,21 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo)
/* hurrah the memory is not visible ! */
atomic64_inc(&adev->num_vram_cpu_page_faults);
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM);
lpfn = adev->mc.visible_vram_size >> PAGE_SHIFT;
for (i = 0; i < abo->placement.num_placement; i++) {
/* Force into visible VRAM */
if ((abo->placements[i].flags & TTM_PL_FLAG_VRAM) &&
(!abo->placements[i].lpfn ||
abo->placements[i].lpfn > lpfn))
abo->placements[i].lpfn = lpfn;
}
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
AMDGPU_GEM_DOMAIN_GTT);
/* Avoid costly evictions; only set GTT as a busy placement */
abo->placement.num_busy_placement = 1;
abo->placement.busy_placement = &abo->placements[1];
r = ttm_bo_validate(bo, &abo->placement, false, false);
if (unlikely(r == -ENOMEM)) {
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
return ttm_bo_validate(bo, &abo->placement, false, false);
} else if (unlikely(r != 0)) {
if (unlikely(r != 0))
return r;
}
offset = bo->mem.start << PAGE_SHIFT;
/* this should never happen */
if ((offset + size) > adev->mc.visible_vram_size)
if (bo->mem.mem_type == TTM_PL_VRAM &&
(offset + size) > adev->mc.visible_vram_size)
return -EINVAL;
return 0;

74
drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
View file

@ -33,6 +33,67 @@
#define AMDGPU_BO_INVALID_OFFSET LONG_MAX
struct amdgpu_bo_va_mapping {
struct list_head list;
struct rb_node rb;
uint64_t start;
uint64_t last;
uint64_t __subtree_last;
uint64_t offset;
uint64_t flags;
};
/* bo virtual addresses in a specific vm */
struct amdgpu_bo_va {
/* protected by bo being reserved */
struct list_head bo_list;
struct dma_fence *last_pt_update;
unsigned ref_count;
/* protected by vm mutex and spinlock */
struct list_head vm_status;
/* mappings for this bo_va */
struct list_head invalids;
struct list_head valids;
/* constant after initialization */
struct amdgpu_vm *vm;
struct amdgpu_bo *bo;
};
struct amdgpu_bo {
/* Protected by tbo.reserved */
u32 preferred_domains;
u32 allowed_domains;
struct ttm_place placements[AMDGPU_GEM_DOMAIN_MAX + 1];
struct ttm_placement placement;
struct ttm_buffer_object tbo;
struct ttm_bo_kmap_obj kmap;
u64 flags;
unsigned pin_count;
u64 tiling_flags;
u64 metadata_flags;
void *metadata;
u32 metadata_size;
unsigned prime_shared_count;
/* list of all virtual address to which this bo is associated to */
struct list_head va;
/* Constant after initialization */
struct drm_gem_object gem_base;
struct amdgpu_bo *parent;
struct amdgpu_bo *shadow;
struct ttm_bo_kmap_obj dma_buf_vmap;
struct amdgpu_mn *mn;
union {
struct list_head mn_list;
struct list_head shadow_list;
};
};
/**
* amdgpu_mem_type_to_domain - return domain corresponding to mem_type
* @mem_type: ttm memory type
@ -120,7 +181,11 @@ static inline u64 amdgpu_bo_mmap_offset(struct amdgpu_bo *bo)
*/
static inline bool amdgpu_bo_gpu_accessible(struct amdgpu_bo *bo)
{
return bo->tbo.mem.mem_type != TTM_PL_SYSTEM;
switch (bo->tbo.mem.mem_type) {
case TTM_PL_TT: return amdgpu_ttm_is_bound(bo->tbo.ttm);
case TTM_PL_VRAM: return true;
default: return false;
}
}
int amdgpu_bo_create(struct amdgpu_device *adev,
@ -128,6 +193,7 @@ int amdgpu_bo_create(struct amdgpu_device *adev,
bool kernel, u32 domain, u64 flags,
struct sg_table *sg,
struct reservation_object *resv,
uint64_t init_value,
struct amdgpu_bo **bo_ptr);
int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
unsigned long size, int byte_align,
@ -135,7 +201,12 @@ int amdgpu_bo_create_restricted(struct amdgpu_device *adev,
struct sg_table *sg,
struct ttm_placement *placement,
struct reservation_object *resv,
uint64_t init_value,
struct amdgpu_bo **bo_ptr);
int amdgpu_bo_create_reserved(struct amdgpu_device *adev,
unsigned long size, int align,
u32 domain, struct amdgpu_bo **bo_ptr,
u64 *gpu_addr, void **cpu_addr);
int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
unsigned long size, int align,
u32 domain, struct amdgpu_bo **bo_ptr,
@ -143,6 +214,7 @@ int amdgpu_bo_create_kernel(struct amdgpu_device *adev,
void amdgpu_bo_free_kernel(struct amdgpu_bo **bo, u64 *gpu_addr,
void **cpu_addr);
int amdgpu_bo_kmap(struct amdgpu_bo *bo, void **ptr);
void *amdgpu_bo_kptr(struct amdgpu_bo *bo);
void amdgpu_bo_kunmap(struct amdgpu_bo *bo);
struct amdgpu_bo *amdgpu_bo_ref(struct amdgpu_bo *bo);
void amdgpu_bo_unref(struct amdgpu_bo **bo);

1
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.h
View file

@ -30,6 +30,7 @@ struct cg_flag_name
const char *name;
};
void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev);
int amdgpu_pm_sysfs_init(struct amdgpu_device *adev);
void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev);
void amdgpu_pm_print_power_states(struct amdgpu_device *adev);

2
drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
View file

@ -69,7 +69,7 @@ amdgpu_gem_prime_import_sg_table(struct drm_device *dev,
ww_mutex_lock(&resv->lock, NULL);
ret = amdgpu_bo_create(adev, attach->dmabuf->size, PAGE_SIZE, false,
AMDGPU_GEM_DOMAIN_GTT, 0, sg, resv, &bo);
AMDGPU_GEM_DOMAIN_GTT, 0, sg, resv, 0, &bo);
ww_mutex_unlock(&resv->lock);
if (ret)
return ERR_PTR(ret);

63
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
View file

@ -63,8 +63,13 @@ static int psp_sw_init(void *handle)
psp->smu_reload_quirk = psp_v3_1_smu_reload_quirk;
break;
case CHIP_RAVEN:
#if 0
psp->init_microcode = psp_v10_0_init_microcode;
#endif
psp->prep_cmd_buf = psp_v10_0_prep_cmd_buf;
psp->ring_init = psp_v10_0_ring_init;
psp->ring_create = psp_v10_0_ring_create;
psp->ring_destroy = psp_v10_0_ring_destroy;
psp->cmd_submit = psp_v10_0_cmd_submit;
psp->compare_sram_data = psp_v10_0_compare_sram_data;
break;
@ -95,9 +100,8 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
int i;
struct amdgpu_device *adev = psp->adev;
val = RREG32(reg_index);
for (i = 0; i < adev->usec_timeout; i++) {
val = RREG32(reg_index);
if (check_changed) {
if (val != reg_val)
return 0;
@ -118,33 +122,18 @@ psp_cmd_submit_buf(struct psp_context *psp,
int index)
{
int ret;
struct amdgpu_bo *cmd_buf_bo;
uint64_t cmd_buf_mc_addr;
struct psp_gfx_cmd_resp *cmd_buf_mem;
struct amdgpu_device *adev = psp->adev;
ret = amdgpu_bo_create_kernel(adev, PSP_CMD_BUFFER_SIZE, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&cmd_buf_bo, &cmd_buf_mc_addr,
(void **)&cmd_buf_mem);
if (ret)
return ret;
memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE);
memset(cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE);
memcpy(psp->cmd_buf_mem, cmd, sizeof(struct psp_gfx_cmd_resp));
memcpy(cmd_buf_mem, cmd, sizeof(struct psp_gfx_cmd_resp));
ret = psp_cmd_submit(psp, ucode, cmd_buf_mc_addr,
ret = psp_cmd_submit(psp, ucode, psp->cmd_buf_mc_addr,
fence_mc_addr, index);
while (*((unsigned int *)psp->fence_buf) != index) {
msleep(1);
}
amdgpu_bo_free_kernel(&cmd_buf_bo,
&cmd_buf_mc_addr,
(void **)&cmd_buf_mem);
return ret;
}
@ -351,6 +340,13 @@ static int psp_load_fw(struct amdgpu_device *adev)
&psp->fence_buf_bo,
&psp->fence_buf_mc_addr,
&psp->fence_buf);
if (ret)
goto failed_mem2;
ret = amdgpu_bo_create_kernel(adev, PSP_CMD_BUFFER_SIZE, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
(void **)&psp->cmd_buf_mem);
if (ret)
goto failed_mem1;
@ -358,7 +354,7 @@ static int psp_load_fw(struct amdgpu_device *adev)
ret = psp_ring_init(psp, PSP_RING_TYPE__KM);
if (ret)
goto failed_mem1;
goto failed_mem;
ret = psp_tmr_init(psp);
if (ret)
@ -379,9 +375,13 @@ static int psp_load_fw(struct amdgpu_device *adev)
return 0;
failed_mem:
amdgpu_bo_free_kernel(&psp->cmd_buf_bo,
&psp->cmd_buf_mc_addr,
(void **)&psp->cmd_buf_mem);
failed_mem1:
amdgpu_bo_free_kernel(&psp->fence_buf_bo,
&psp->fence_buf_mc_addr, &psp->fence_buf);
failed_mem1:
failed_mem2:
amdgpu_bo_free_kernel(&psp->fw_pri_bo,
&psp->fw_pri_mc_addr, &psp->fw_pri_buf);
failed:
@ -435,16 +435,15 @@ static int psp_hw_fini(void *handle)
psp_ring_destroy(psp, PSP_RING_TYPE__KM);
if (psp->tmr_buf)
amdgpu_bo_free_kernel(&psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
if (psp->fw_pri_buf)
amdgpu_bo_free_kernel(&psp->fw_pri_bo,
&psp->fw_pri_mc_addr, &psp->fw_pri_buf);
if (psp->fence_buf_bo)
amdgpu_bo_free_kernel(&psp->fence_buf_bo,
&psp->fence_buf_mc_addr, &psp->fence_buf);
amdgpu_bo_free_kernel(&psp->tmr_bo, &psp->tmr_mc_addr, &psp->tmr_buf);
amdgpu_bo_free_kernel(&psp->fw_pri_bo,
&psp->fw_pri_mc_addr, &psp->fw_pri_buf);
amdgpu_bo_free_kernel(&psp->fence_buf_bo,
&psp->fence_buf_mc_addr, &psp->fence_buf);
amdgpu_bo_free_kernel(&psp->asd_shared_bo, &psp->asd_shared_mc_addr,
&psp->asd_shared_buf);
amdgpu_bo_free_kernel(&psp->cmd_buf_bo, &psp->cmd_buf_mc_addr,
(void **)&psp->cmd_buf_mem);
kfree(psp->cmd);
psp->cmd = NULL;

5
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
View file

@ -108,6 +108,11 @@ struct psp_context
struct amdgpu_bo *fence_buf_bo;
uint64_t fence_buf_mc_addr;
void *fence_buf;
/* cmd buffer */
struct amdgpu_bo *cmd_buf_bo;
uint64_t cmd_buf_mc_addr;
struct psp_gfx_cmd_resp *cmd_buf_mem;
};
struct amdgpu_psp_funcs {

53
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
View file

@ -184,32 +184,16 @@ int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
return r;
}
if (ring->funcs->support_64bit_ptrs) {
r = amdgpu_wb_get_64bit(adev, &ring->rptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get_64bit(adev, &ring->wptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
return r;
}
} else {
r = amdgpu_wb_get(adev, &ring->rptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->wptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->rptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring rptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->wptr_offs);
if (r) {
dev_err(adev->dev, "(%d) ring wptr_offs wb alloc failed\n", r);
return r;
}
r = amdgpu_wb_get(adev, &ring->fence_offs);
@ -277,18 +261,15 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
{
ring->ready = false;
if (ring->funcs->support_64bit_ptrs) {
amdgpu_wb_free_64bit(ring->adev, ring->cond_exe_offs);
amdgpu_wb_free_64bit(ring->adev, ring->fence_offs);
amdgpu_wb_free_64bit(ring->adev, ring->rptr_offs);
amdgpu_wb_free_64bit(ring->adev, ring->wptr_offs);
} else {
amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
amdgpu_wb_free(ring->adev, ring->fence_offs);
amdgpu_wb_free(ring->adev, ring->rptr_offs);
amdgpu_wb_free(ring->adev, ring->wptr_offs);
}
/* Not to finish a ring which is not initialized */
if (!(ring->adev) || !(ring->adev->rings[ring->idx]))
return;
amdgpu_wb_free(ring->adev, ring->rptr_offs);
amdgpu_wb_free(ring->adev, ring->wptr_offs);
amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
amdgpu_wb_free(ring->adev, ring->fence_offs);
amdgpu_bo_free_kernel(&ring->ring_obj,
&ring->gpu_addr,

40
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
View file

@ -212,4 +212,44 @@ static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
}
static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
{
if (ring->count_dw <= 0)
DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
ring->ring[ring->wptr++ & ring->buf_mask] = v;
ring->wptr &= ring->ptr_mask;
ring->count_dw--;
}
static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
void *src, int count_dw)
{
unsigned occupied, chunk1, chunk2;
void *dst;
if (unlikely(ring->count_dw < count_dw))
DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
occupied = ring->wptr & ring->buf_mask;
dst = (void *)&ring->ring[occupied];
chunk1 = ring->buf_mask + 1 - occupied;
chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
chunk2 = count_dw - chunk1;
chunk1 <<= 2;
chunk2 <<= 2;
if (chunk1)
memcpy(dst, src, chunk1);
if (chunk2) {
src += chunk1;
dst = (void *)ring->ring;
memcpy(dst, src, chunk2);
}
ring->wptr += count_dw;
ring->wptr &= ring->ptr_mask;
ring->count_dw -= count_dw;
}
#endif

2
drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
View file

@ -64,7 +64,7 @@ int amdgpu_sa_bo_manager_init(struct amdgpu_device *adev,
INIT_LIST_HEAD(&sa_manager->flist[i]);
r = amdgpu_bo_create(adev, size, align, true, domain,
0, NULL, NULL, &sa_manager->bo);
0, NULL, NULL, 0, &sa_manager->bo);
if (r) {
dev_err(adev->dev, "(%d) failed to allocate bo for manager\n", r);
return r;

56
drivers/gpu/drm/amd/amdgpu/amdgpu_test.c
View file

@ -33,7 +33,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
struct amdgpu_bo *vram_obj = NULL;
struct amdgpu_bo **gtt_obj = NULL;
uint64_t gtt_addr, vram_addr;
uint64_t gart_addr, vram_addr;
unsigned n, size;
int i, r;
@ -42,7 +42,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
/* Number of tests =
* (Total GTT - IB pool - writeback page - ring buffers) / test size
*/
n = adev->mc.gtt_size - AMDGPU_IB_POOL_SIZE*64*1024;
n = adev->mc.gart_size - AMDGPU_IB_POOL_SIZE*64*1024;
for (i = 0; i < AMDGPU_MAX_RINGS; ++i)
if (adev->rings[i])
n -= adev->rings[i]->ring_size;
@ -61,7 +61,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM, 0,
NULL, NULL, &vram_obj);
NULL, NULL, 0, &vram_obj);
if (r) {
DRM_ERROR("Failed to create VRAM object\n");
goto out_cleanup;
@ -76,13 +76,13 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
}
for (i = 0; i < n; i++) {
void *gtt_map, *vram_map;
void **gtt_start, **gtt_end;
void **gart_start, **gart_end;
void **vram_start, **vram_end;
struct dma_fence *fence = NULL;
r = amdgpu_bo_create(adev, size, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0, NULL,
NULL, gtt_obj + i);
NULL, 0, gtt_obj + i);
if (r) {
DRM_ERROR("Failed to create GTT object %d\n", i);
goto out_lclean;
@ -91,7 +91,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
r = amdgpu_bo_reserve(gtt_obj[i], false);
if (unlikely(r != 0))
goto out_lclean_unref;
r = amdgpu_bo_pin(gtt_obj[i], AMDGPU_GEM_DOMAIN_GTT, &gtt_addr);
r = amdgpu_bo_pin(gtt_obj[i], AMDGPU_GEM_DOMAIN_GTT, &gart_addr);
if (r) {
DRM_ERROR("Failed to pin GTT object %d\n", i);
goto out_lclean_unres;
@ -103,15 +103,15 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
goto out_lclean_unpin;
}
for (gtt_start = gtt_map, gtt_end = gtt_map + size;
gtt_start < gtt_end;
gtt_start++)
*gtt_start = gtt_start;
for (gart_start = gtt_map, gart_end = gtt_map + size;
gart_start < gart_end;
gart_start++)
*gart_start = gart_start;
amdgpu_bo_kunmap(gtt_obj[i]);
r = amdgpu_copy_buffer(ring, gtt_addr, vram_addr,
size, NULL, &fence, false);
r = amdgpu_copy_buffer(ring, gart_addr, vram_addr,
size, NULL, &fence, false, false);
if (r) {
DRM_ERROR("Failed GTT->VRAM copy %d\n", i);
@ -132,21 +132,21 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
goto out_lclean_unpin;
}
for (gtt_start = gtt_map, gtt_end = gtt_map + size,
for (gart_start = gtt_map, gart_end = gtt_map + size,
vram_start = vram_map, vram_end = vram_map + size;
vram_start < vram_end;
gtt_start++, vram_start++) {
if (*vram_start != gtt_start) {
gart_start++, vram_start++) {
if (*vram_start != gart_start) {
DRM_ERROR("Incorrect GTT->VRAM copy %d: Got 0x%p, "
"expected 0x%p (GTT/VRAM offset "
"0x%16llx/0x%16llx)\n",
i, *vram_start, gtt_start,
i, *vram_start, gart_start,
(unsigned long long)
(gtt_addr - adev->mc.gtt_start +
(void*)gtt_start - gtt_map),
(gart_addr - adev->mc.gart_start +
(void*)gart_start - gtt_map),
(unsigned long long)
(vram_addr - adev->mc.vram_start +
(void*)gtt_start - gtt_map));
(void*)gart_start - gtt_map));
amdgpu_bo_kunmap(vram_obj);
goto out_lclean_unpin;
}
@ -155,8 +155,8 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
amdgpu_bo_kunmap(vram_obj);
r = amdgpu_copy_buffer(ring, vram_addr, gtt_addr,
size, NULL, &fence, false);
r = amdgpu_copy_buffer(ring, vram_addr, gart_addr,
size, NULL, &fence, false, false);
if (r) {
DRM_ERROR("Failed VRAM->GTT copy %d\n", i);
@ -177,20 +177,20 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
goto out_lclean_unpin;
}
for (gtt_start = gtt_map, gtt_end = gtt_map + size,
for (gart_start = gtt_map, gart_end = gtt_map + size,
vram_start = vram_map, vram_end = vram_map + size;
gtt_start < gtt_end;
gtt_start++, vram_start++) {
if (*gtt_start != vram_start) {
gart_start < gart_end;
gart_start++, vram_start++) {
if (*gart_start != vram_start) {
DRM_ERROR("Incorrect VRAM->GTT copy %d: Got 0x%p, "
"expected 0x%p (VRAM/GTT offset "
"0x%16llx/0x%16llx)\n",
i, *gtt_start, vram_start,
i, *gart_start, vram_start,
(unsigned long long)
(vram_addr - adev->mc.vram_start +
(void*)vram_start - vram_map),
(unsigned long long)
(gtt_addr - adev->mc.gtt_start +
(gart_addr - adev->mc.gart_start +
(void*)vram_start - vram_map));
amdgpu_bo_kunmap(gtt_obj[i]);
goto out_lclean_unpin;
@ -200,7 +200,7 @@ static void amdgpu_do_test_moves(struct amdgpu_device *adev)
amdgpu_bo_kunmap(gtt_obj[i]);
DRM_INFO("Tested GTT->VRAM and VRAM->GTT copy for GTT offset 0x%llx\n",
gtt_addr - adev->mc.gtt_start);
gart_addr - adev->mc.gart_start);
continue;
out_lclean_unpin:

22
drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
View file

@ -105,12 +105,12 @@ TRACE_EVENT(amdgpu_bo_create,
__entry->bo = bo;
__entry->pages = bo->tbo.num_pages;
__entry->type = bo->tbo.mem.mem_type;
__entry->prefer = bo->prefered_domains;
__entry->prefer = bo->preferred_domains;
__entry->allow = bo->allowed_domains;
__entry->visible = bo->flags;
),
TP_printk("bo=%p, pages=%u, type=%d, prefered=%d, allowed=%d, visible=%d",
TP_printk("bo=%p, pages=%u, type=%d, preferred=%d, allowed=%d, visible=%d",
__entry->bo, __entry->pages, __entry->type,
__entry->prefer, __entry->allow, __entry->visible)
);
@ -224,7 +224,7 @@ TRACE_EVENT(amdgpu_vm_bo_map,
__field(long, start)
__field(long, last)
__field(u64, offset)
__field(u32, flags)
__field(u64, flags)
),
TP_fast_assign(
@ -234,7 +234,7 @@ TRACE_EVENT(amdgpu_vm_bo_map,
__entry->offset = mapping->offset;
__entry->flags = mapping->flags;
),
TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%08x",
TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%llx",
__entry->bo, __entry->start, __entry->last,
__entry->offset, __entry->flags)
);
@ -248,7 +248,7 @@ TRACE_EVENT(amdgpu_vm_bo_unmap,
__field(long, start)
__field(long, last)
__field(u64, offset)
__field(u32, flags)
__field(u64, flags)
),
TP_fast_assign(
@ -258,7 +258,7 @@ TRACE_EVENT(amdgpu_vm_bo_unmap,
__entry->offset = mapping->offset;
__entry->flags = mapping->flags;
),
TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%08x",
TP_printk("bo=%p, start=%lx, last=%lx, offset=%010llx, flags=%llx",
__entry->bo, __entry->start, __entry->last,
__entry->offset, __entry->flags)
);
@ -269,7 +269,7 @@ DECLARE_EVENT_CLASS(amdgpu_vm_mapping,
TP_STRUCT__entry(
__field(u64, soffset)
__field(u64, eoffset)
__field(u32, flags)
__field(u64, flags)
),
TP_fast_assign(
@ -277,7 +277,7 @@ DECLARE_EVENT_CLASS(amdgpu_vm_mapping,
__entry->eoffset = mapping->last + 1;
__entry->flags = mapping->flags;
),
TP_printk("soffs=%010llx, eoffs=%010llx, flags=%08x",
TP_printk("soffs=%010llx, eoffs=%010llx, flags=%llx",
__entry->soffset, __entry->eoffset, __entry->flags)
);
@ -293,14 +293,14 @@ DEFINE_EVENT(amdgpu_vm_mapping, amdgpu_vm_bo_mapping,
TRACE_EVENT(amdgpu_vm_set_ptes,
TP_PROTO(uint64_t pe, uint64_t addr, unsigned count,
uint32_t incr, uint32_t flags),
uint32_t incr, uint64_t flags),
TP_ARGS(pe, addr, count, incr, flags),
TP_STRUCT__entry(
__field(u64, pe)
__field(u64, addr)
__field(u32, count)
__field(u32, incr)
__field(u32, flags)
__field(u64, flags)
),
TP_fast_assign(
@ -310,7 +310,7 @@ TRACE_EVENT(amdgpu_vm_set_ptes,
__entry->incr = incr;
__entry->flags = flags;
),
TP_printk("pe=%010Lx, addr=%010Lx, incr=%u, flags=%08x, count=%u",
TP_printk("pe=%010Lx, addr=%010Lx, incr=%u, flags=%llx, count=%u",
__entry->pe, __entry->addr, __entry->incr,
__entry->flags, __entry->count)
);

412
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
View file

@ -47,10 +47,15 @@
#define DRM_FILE_PAGE_OFFSET (0x100000000ULL >> PAGE_SHIFT)
static int amdgpu_map_buffer(struct ttm_buffer_object *bo,
struct ttm_mem_reg *mem, unsigned num_pages,
uint64_t offset, unsigned window,
struct amdgpu_ring *ring,
uint64_t *addr);
static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev);
static void amdgpu_ttm_debugfs_fini(struct amdgpu_device *adev);
/*
* Global memory.
*/
@ -97,6 +102,8 @@ static int amdgpu_ttm_global_init(struct amdgpu_device *adev)
goto error_bo;
}
mutex_init(&adev->mman.gtt_window_lock);
ring = adev->mman.buffer_funcs_ring;
rq = &ring->sched.sched_rq[AMD_SCHED_PRIORITY_KERNEL];
r = amd_sched_entity_init(&ring->sched, &adev->mman.entity,
@ -123,6 +130,7 @@ static void amdgpu_ttm_global_fini(struct amdgpu_device *adev)
if (adev->mman.mem_global_referenced) {
amd_sched_entity_fini(adev->mman.entity.sched,
&adev->mman.entity);
mutex_destroy(&adev->mman.gtt_window_lock);
drm_global_item_unref(&adev->mman.bo_global_ref.ref);
drm_global_item_unref(&adev->mman.mem_global_ref);
adev->mman.mem_global_referenced = false;
@ -150,7 +158,7 @@ static int amdgpu_init_mem_type(struct ttm_bo_device *bdev, uint32_t type,
break;
case TTM_PL_TT:
man->func = &amdgpu_gtt_mgr_func;
man->gpu_offset = adev->mc.gtt_start;
man->gpu_offset = adev->mc.gart_start;
man->available_caching = TTM_PL_MASK_CACHING;
man->default_caching = TTM_PL_FLAG_CACHED;
man->flags = TTM_MEMTYPE_FLAG_MAPPABLE | TTM_MEMTYPE_FLAG_CMA;
@ -186,12 +194,11 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
{
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->bdev);
struct amdgpu_bo *abo;
static struct ttm_place placements = {
static const struct ttm_place placements = {
.fpfn = 0,
.lpfn = 0,
.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_SYSTEM
};
unsigned i;
if (!amdgpu_ttm_bo_is_amdgpu_bo(bo)) {
placement->placement = &placements;
@ -207,22 +214,36 @@ static void amdgpu_evict_flags(struct ttm_buffer_object *bo,
adev->mman.buffer_funcs_ring &&
adev->mman.buffer_funcs_ring->ready == false) {
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_CPU);
} else {
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
for (i = 0; i < abo->placement.num_placement; ++i) {
if (!(abo->placements[i].flags &
TTM_PL_FLAG_TT))
continue;
} else if (adev->mc.visible_vram_size < adev->mc.real_vram_size &&
!(abo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)) {
unsigned fpfn = adev->mc.visible_vram_size >> PAGE_SHIFT;
struct drm_mm_node *node = bo->mem.mm_node;
unsigned long pages_left;
if (abo->placements[i].lpfn)
continue;
/* set an upper limit to force directly
* allocating address space for the BO.
*/
abo->placements[i].lpfn =
adev->mc.gtt_size >> PAGE_SHIFT;
for (pages_left = bo->mem.num_pages;
pages_left;
pages_left -= node->size, node++) {
if (node->start < fpfn)
break;
}
if (!pages_left)
goto gtt;
/* Try evicting to the CPU inaccessible part of VRAM
* first, but only set GTT as busy placement, so this
* BO will be evicted to GTT rather than causing other
* BOs to be evicted from VRAM
*/
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_VRAM |
AMDGPU_GEM_DOMAIN_GTT);
abo->placements[0].fpfn = fpfn;
abo->placements[0].lpfn = 0;
abo->placement.busy_placement = &abo->placements[1];
abo->placement.num_busy_placement = 1;
} else {
gtt:
amdgpu_ttm_placement_from_domain(abo, AMDGPU_GEM_DOMAIN_GTT);
}
break;
case TTM_PL_TT:
@ -252,29 +273,18 @@ static void amdgpu_move_null(struct ttm_buffer_object *bo,
new_mem->mm_node = NULL;
}
static int amdgpu_mm_node_addr(struct ttm_buffer_object *bo,
struct drm_mm_node *mm_node,
struct ttm_mem_reg *mem,
uint64_t *addr)
static uint64_t amdgpu_mm_node_addr(struct ttm_buffer_object *bo,
struct drm_mm_node *mm_node,
struct ttm_mem_reg *mem)
{
int r;
uint64_t addr = 0;
switch (mem->mem_type) {
case TTM_PL_TT:
r = amdgpu_ttm_bind(bo, mem);
if (r)
return r;
case TTM_PL_VRAM:
*addr = mm_node->start << PAGE_SHIFT;
*addr += bo->bdev->man[mem->mem_type].gpu_offset;
break;
default:
DRM_ERROR("Unknown placement %d\n", mem->mem_type);
return -EINVAL;
if (mem->mem_type != TTM_PL_TT ||
amdgpu_gtt_mgr_is_allocated(mem)) {
addr = mm_node->start << PAGE_SHIFT;
addr += bo->bdev->man[mem->mem_type].gpu_offset;
}
return 0;
return addr;
}
static int amdgpu_move_blit(struct ttm_buffer_object *bo,
@ -299,26 +309,40 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
}
old_mm = old_mem->mm_node;
r = amdgpu_mm_node_addr(bo, old_mm, old_mem, &old_start);
if (r)
return r;
old_size = old_mm->size;
old_start = amdgpu_mm_node_addr(bo, old_mm, old_mem);
new_mm = new_mem->mm_node;
r = amdgpu_mm_node_addr(bo, new_mm, new_mem, &new_start);
if (r)
return r;
new_size = new_mm->size;
new_start = amdgpu_mm_node_addr(bo, new_mm, new_mem);
num_pages = new_mem->num_pages;
mutex_lock(&adev->mman.gtt_window_lock);
while (num_pages) {
unsigned long cur_pages = min(old_size, new_size);
unsigned long cur_pages = min(min(old_size, new_size),
(u64)AMDGPU_GTT_MAX_TRANSFER_SIZE);
uint64_t from = old_start, to = new_start;
struct dma_fence *next;
r = amdgpu_copy_buffer(ring, old_start, new_start,
if (old_mem->mem_type == TTM_PL_TT &&
!amdgpu_gtt_mgr_is_allocated(old_mem)) {
r = amdgpu_map_buffer(bo, old_mem, cur_pages,
old_start, 0, ring, &from);
if (r)
goto error;
}
if (new_mem->mem_type == TTM_PL_TT &&
!amdgpu_gtt_mgr_is_allocated(new_mem)) {
r = amdgpu_map_buffer(bo, new_mem, cur_pages,
new_start, 1, ring, &to);
if (r)
goto error;
}
r = amdgpu_copy_buffer(ring, from, to,
cur_pages * PAGE_SIZE,
bo->resv, &next, false);
bo->resv, &next, false, true);
if (r)
goto error;
@ -331,10 +355,7 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
old_size -= cur_pages;
if (!old_size) {
r = amdgpu_mm_node_addr(bo, ++old_mm, old_mem,
&old_start);
if (r)
goto error;
old_start = amdgpu_mm_node_addr(bo, ++old_mm, old_mem);
old_size = old_mm->size;
} else {
old_start += cur_pages * PAGE_SIZE;
@ -342,22 +363,21 @@ static int amdgpu_move_blit(struct ttm_buffer_object *bo,
new_size -= cur_pages;
if (!new_size) {
r = amdgpu_mm_node_addr(bo, ++new_mm, new_mem,
&new_start);
if (r)
goto error;
new_start = amdgpu_mm_node_addr(bo, ++new_mm, new_mem);
new_size = new_mm->size;
} else {
new_start += cur_pages * PAGE_SIZE;
}
}
mutex_unlock(&adev->mman.gtt_window_lock);
r = ttm_bo_pipeline_move(bo, fence, evict, new_mem);
dma_fence_put(fence);
return r;
error:
mutex_unlock(&adev->mman.gtt_window_lock);
if (fence)
dma_fence_wait(fence, false);
dma_fence_put(fence);
@ -384,7 +404,7 @@ static int amdgpu_move_vram_ram(struct ttm_buffer_object *bo,
placement.num_busy_placement = 1;
placement.busy_placement = &placements;
placements.fpfn = 0;
placements.lpfn = adev->mc.gtt_size >> PAGE_SHIFT;
placements.lpfn = 0;
placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
interruptible, no_wait_gpu);
@ -431,7 +451,7 @@ static int amdgpu_move_ram_vram(struct ttm_buffer_object *bo,
placement.num_busy_placement = 1;
placement.busy_placement = &placements;
placements.fpfn = 0;
placements.lpfn = adev->mc.gtt_size >> PAGE_SHIFT;
placements.lpfn = 0;
placements.flags = TTM_PL_MASK_CACHING | TTM_PL_FLAG_TT;
r = ttm_bo_mem_space(bo, &placement, &tmp_mem,
interruptible, no_wait_gpu);
@ -507,6 +527,15 @@ memcpy:
}
}
if (bo->type == ttm_bo_type_device &&
new_mem->mem_type == TTM_PL_VRAM &&
old_mem->mem_type != TTM_PL_VRAM) {
/* amdgpu_bo_fault_reserve_notify will re-set this if the CPU
* accesses the BO after it's moved.
*/
abo->flags &= ~AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED;
}
/* update statistics */
atomic64_add((u64)bo->num_pages << PAGE_SHIFT, &adev->num_bytes_moved);
return 0;
@ -695,11 +724,36 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_tt *ttm)
sg_free_table(ttm->sg);
}
static int amdgpu_ttm_do_bind(struct ttm_tt *ttm, struct ttm_mem_reg *mem)
{
struct amdgpu_ttm_tt *gtt = (void *)ttm;
uint64_t flags;
int r;
spin_lock(&gtt->adev->gtt_list_lock);
flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, mem);
gtt->offset = (u64)mem->start << PAGE_SHIFT;
r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
ttm->pages, gtt->ttm.dma_address, flags);
if (r) {
DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
ttm->num_pages, gtt->offset);
goto error_gart_bind;
}
list_add_tail(&gtt->list, &gtt->adev->gtt_list);
error_gart_bind:
spin_unlock(&gtt->adev->gtt_list_lock);
return r;
}
static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
struct ttm_mem_reg *bo_mem)
{
struct amdgpu_ttm_tt *gtt = (void*)ttm;
int r;
int r = 0;
if (gtt->userptr) {
r = amdgpu_ttm_tt_pin_userptr(ttm);
@ -718,7 +772,10 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
bo_mem->mem_type == AMDGPU_PL_OA)
return -EINVAL;
return 0;
if (amdgpu_gtt_mgr_is_allocated(bo_mem))
r = amdgpu_ttm_do_bind(ttm, bo_mem);
return r;
}
bool amdgpu_ttm_is_bound(struct ttm_tt *ttm)
@ -731,8 +788,6 @@ bool amdgpu_ttm_is_bound(struct ttm_tt *ttm)
int amdgpu_ttm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *bo_mem)
{
struct ttm_tt *ttm = bo->ttm;
struct amdgpu_ttm_tt *gtt = (void *)bo->ttm;
uint64_t flags;
int r;
if (!ttm || amdgpu_ttm_is_bound(ttm))
@ -745,22 +800,7 @@ int amdgpu_ttm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *bo_mem)
return r;
}
spin_lock(&gtt->adev->gtt_list_lock);
flags = amdgpu_ttm_tt_pte_flags(gtt->adev, ttm, bo_mem);
gtt->offset = (u64)bo_mem->start << PAGE_SHIFT;
r = amdgpu_gart_bind(gtt->adev, gtt->offset, ttm->num_pages,
ttm->pages, gtt->ttm.dma_address, flags);
if (r) {
DRM_ERROR("failed to bind %lu pages at 0x%08llX\n",
ttm->num_pages, gtt->offset);
goto error_gart_bind;
}
list_add_tail(&gtt->list, &gtt->adev->gtt_list);
error_gart_bind:
spin_unlock(&gtt->adev->gtt_list_lock);
return r;
return amdgpu_ttm_do_bind(ttm, bo_mem);
}
int amdgpu_ttm_recover_gart(struct amdgpu_device *adev)
@ -1075,6 +1115,67 @@ static bool amdgpu_ttm_bo_eviction_valuable(struct ttm_buffer_object *bo,
return ttm_bo_eviction_valuable(bo, place);
}
static int amdgpu_ttm_access_memory(struct ttm_buffer_object *bo,
unsigned long offset,
void *buf, int len, int write)
{
struct amdgpu_bo *abo = container_of(bo, struct amdgpu_bo, tbo);
struct amdgpu_device *adev = amdgpu_ttm_adev(abo->tbo.bdev);
struct drm_mm_node *nodes = abo->tbo.mem.mm_node;
uint32_t value = 0;
int ret = 0;
uint64_t pos;
unsigned long flags;
if (bo->mem.mem_type != TTM_PL_VRAM)
return -EIO;
while (offset >= (nodes->size << PAGE_SHIFT)) {
offset -= nodes->size << PAGE_SHIFT;
++nodes;
}
pos = (nodes->start << PAGE_SHIFT) + offset;
while (len && pos < adev->mc.mc_vram_size) {
uint64_t aligned_pos = pos & ~(uint64_t)3;
uint32_t bytes = 4 - (pos & 3);
uint32_t shift = (pos & 3) * 8;
uint32_t mask = 0xffffffff << shift;
if (len < bytes) {
mask &= 0xffffffff >> (bytes - len) * 8;
bytes = len;
}
spin_lock_irqsave(&adev->mmio_idx_lock, flags);
WREG32(mmMM_INDEX, ((uint32_t)aligned_pos) | 0x80000000);
WREG32(mmMM_INDEX_HI, aligned_pos >> 31);
if (!write || mask != 0xffffffff)
value = RREG32(mmMM_DATA);
if (write) {
value &= ~mask;
value |= (*(uint32_t *)buf << shift) & mask;
WREG32(mmMM_DATA, value);
}
spin_unlock_irqrestore(&adev->mmio_idx_lock, flags);
if (!write) {
value = (value & mask) >> shift;
memcpy(buf, &value, bytes);
}
ret += bytes;
buf = (uint8_t *)buf + bytes;
pos += bytes;
len -= bytes;
if (pos >= (nodes->start + nodes->size) << PAGE_SHIFT) {
++nodes;
pos = (nodes->start << PAGE_SHIFT);
}
}
return ret;
}
static struct ttm_bo_driver amdgpu_bo_driver = {
.ttm_tt_create = &amdgpu_ttm_tt_create,
.ttm_tt_populate = &amdgpu_ttm_tt_populate,
@ -1090,11 +1191,14 @@ static struct ttm_bo_driver amdgpu_bo_driver = {
.io_mem_reserve = &amdgpu_ttm_io_mem_reserve,
.io_mem_free = &amdgpu_ttm_io_mem_free,
.io_mem_pfn = amdgpu_ttm_io_mem_pfn,
.access_memory = &amdgpu_ttm_access_memory
};
int amdgpu_ttm_init(struct amdgpu_device *adev)
{
uint64_t gtt_size;
int r;
u64 vis_vram_limit;
r = amdgpu_ttm_global_init(adev);
if (r) {
@ -1118,36 +1222,37 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
DRM_ERROR("Failed initializing VRAM heap.\n");
return r;
}
/* Reduce size of CPU-visible VRAM if requested */
vis_vram_limit = (u64)amdgpu_vis_vram_limit * 1024 * 1024;
if (amdgpu_vis_vram_limit > 0 &&
vis_vram_limit <= adev->mc.visible_vram_size)
adev->mc.visible_vram_size = vis_vram_limit;
/* Change the size here instead of the init above so only lpfn is affected */
amdgpu_ttm_set_active_vram_size(adev, adev->mc.visible_vram_size);
r = amdgpu_bo_create(adev, adev->mc.stolen_size, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &adev->stollen_vga_memory);
if (r) {
return r;
}
r = amdgpu_bo_reserve(adev->stollen_vga_memory, false);
r = amdgpu_bo_create_kernel(adev, adev->mc.stolen_size, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->stolen_vga_memory,
NULL, NULL);
if (r)
return r;
r = amdgpu_bo_pin(adev->stollen_vga_memory, AMDGPU_GEM_DOMAIN_VRAM, NULL);
amdgpu_bo_unreserve(adev->stollen_vga_memory);
if (r) {
amdgpu_bo_unref(&adev->stollen_vga_memory);
return r;
}
DRM_INFO("amdgpu: %uM of VRAM memory ready\n",
(unsigned) (adev->mc.real_vram_size / (1024 * 1024)));
r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_TT,
adev->mc.gtt_size >> PAGE_SHIFT);
if (amdgpu_gtt_size == -1)
gtt_size = max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
adev->mc.mc_vram_size);
else
gtt_size = (uint64_t)amdgpu_gtt_size << 20;
r = ttm_bo_init_mm(&adev->mman.bdev, TTM_PL_TT, gtt_size >> PAGE_SHIFT);
if (r) {
DRM_ERROR("Failed initializing GTT heap.\n");
return r;
}
DRM_INFO("amdgpu: %uM of GTT memory ready.\n",
(unsigned)(adev->mc.gtt_size / (1024 * 1024)));
(unsigned)(gtt_size / (1024 * 1024)));
adev->gds.mem.total_size = adev->gds.mem.total_size << AMDGPU_GDS_SHIFT;
adev->gds.mem.gfx_partition_size = adev->gds.mem.gfx_partition_size << AMDGPU_GDS_SHIFT;
@ -1203,13 +1308,13 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev)
if (!adev->mman.initialized)
return;
amdgpu_ttm_debugfs_fini(adev);
if (adev->stollen_vga_memory) {
r = amdgpu_bo_reserve(adev->stollen_vga_memory, true);
if (adev->stolen_vga_memory) {
r = amdgpu_bo_reserve(adev->stolen_vga_memory, true);
if (r == 0) {
amdgpu_bo_unpin(adev->stollen_vga_memory);
amdgpu_bo_unreserve(adev->stollen_vga_memory);
amdgpu_bo_unpin(adev->stolen_vga_memory);
amdgpu_bo_unreserve(adev->stolen_vga_memory);
}
amdgpu_bo_unref(&adev->stollen_vga_memory);
amdgpu_bo_unref(&adev->stolen_vga_memory);
}
ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_VRAM);
ttm_bo_clean_mm(&adev->mman.bdev, TTM_PL_TT);
@ -1256,12 +1361,77 @@ int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma)
return ttm_bo_mmap(filp, vma, &adev->mman.bdev);
}
int amdgpu_copy_buffer(struct amdgpu_ring *ring,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
static int amdgpu_map_buffer(struct ttm_buffer_object *bo,
struct ttm_mem_reg *mem, unsigned num_pages,
uint64_t offset, unsigned window,
struct amdgpu_ring *ring,
uint64_t *addr)
{
struct amdgpu_ttm_tt *gtt = (void *)bo->ttm;
struct amdgpu_device *adev = ring->adev;
struct ttm_tt *ttm = bo->ttm;
struct amdgpu_job *job;
unsigned num_dw, num_bytes;
dma_addr_t *dma_address;
struct dma_fence *fence;
uint64_t src_addr, dst_addr;
uint64_t flags;
int r;
BUG_ON(adev->mman.buffer_funcs->copy_max_bytes <
AMDGPU_GTT_MAX_TRANSFER_SIZE * 8);
*addr = adev->mc.gart_start;
*addr += (u64)window * AMDGPU_GTT_MAX_TRANSFER_SIZE *
AMDGPU_GPU_PAGE_SIZE;
num_dw = adev->mman.buffer_funcs->copy_num_dw;
while (num_dw & 0x7)
num_dw++;
num_bytes = num_pages * 8;
r = amdgpu_job_alloc_with_ib(adev, num_dw * 4 + num_bytes, &job);
if (r)
return r;
src_addr = num_dw * 4;
src_addr += job->ibs[0].gpu_addr;
dst_addr = adev->gart.table_addr;
dst_addr += window * AMDGPU_GTT_MAX_TRANSFER_SIZE * 8;
amdgpu_emit_copy_buffer(adev, &job->ibs[0], src_addr,
dst_addr, num_bytes);
amdgpu_ring_pad_ib(ring, &job->ibs[0]);
WARN_ON(job->ibs[0].length_dw > num_dw);
dma_address = &gtt->ttm.dma_address[offset >> PAGE_SHIFT];
flags = amdgpu_ttm_tt_pte_flags(adev, ttm, mem);
r = amdgpu_gart_map(adev, 0, num_pages, dma_address, flags,
&job->ibs[0].ptr[num_dw]);
if (r)
goto error_free;
r = amdgpu_job_submit(job, ring, &adev->mman.entity,
AMDGPU_FENCE_OWNER_UNDEFINED, &fence);
if (r)
goto error_free;
dma_fence_put(fence);
return r;
error_free:
amdgpu_job_free(job);
return r;
}
int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
uint64_t dst_offset, uint32_t byte_count,
struct reservation_object *resv,
struct dma_fence **fence, bool direct_submit)
struct dma_fence **fence, bool direct_submit,
bool vm_needs_flush)
{
struct amdgpu_device *adev = ring->adev;
struct amdgpu_job *job;
@ -1283,6 +1453,7 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring,
if (r)
return r;
job->vm_needs_flush = vm_needs_flush;
if (resv) {
r = amdgpu_sync_resv(adev, &job->sync, resv,
AMDGPU_FENCE_OWNER_UNDEFINED);
@ -1327,11 +1498,12 @@ error_free:
}
int amdgpu_fill_buffer(struct amdgpu_bo *bo,
uint32_t src_data,
uint64_t src_data,
struct reservation_object *resv,
struct dma_fence **fence)
{
struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
/* max_bytes applies to SDMA_OP_PTEPDE as well as SDMA_OP_CONST_FILL*/
uint32_t max_bytes = adev->mman.buffer_funcs->fill_max_bytes;
struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring;
@ -1347,6 +1519,12 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
return -EINVAL;
}
if (bo->tbo.mem.mem_type == TTM_PL_TT) {
r = amdgpu_ttm_bind(&bo->tbo, &bo->tbo.mem);
if (r)
return r;
}
num_pages = bo->tbo.num_pages;
mm_node = bo->tbo.mem.mm_node;
num_loops = 0;
@ -1357,7 +1535,9 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
num_pages -= mm_node->size;
++mm_node;
}
num_dw = num_loops * adev->mman.buffer_funcs->fill_num_dw;
/* 10 double words for each SDMA_OP_PTEPDE cmd */
num_dw = num_loops * 10;
/* for IB padding */
num_dw += 64;
@ -1382,16 +1562,16 @@ int amdgpu_fill_buffer(struct amdgpu_bo *bo,
uint32_t byte_count = mm_node->size << PAGE_SHIFT;
uint64_t dst_addr;
r = amdgpu_mm_node_addr(&bo->tbo, mm_node,
&bo->tbo.mem, &dst_addr);
if (r)
return r;
WARN_ONCE(byte_count & 0x7, "size should be a multiple of 8");
dst_addr = amdgpu_mm_node_addr(&bo->tbo, mm_node, &bo->tbo.mem);
while (byte_count) {
uint32_t cur_size_in_bytes = min(byte_count, max_bytes);
amdgpu_emit_fill_buffer(adev, &job->ibs[0], src_data,
dst_addr, cur_size_in_bytes);
amdgpu_vm_set_pte_pde(adev, &job->ibs[0],
dst_addr, 0,
cur_size_in_bytes >> 3, 0,
src_data);
dst_addr += cur_size_in_bytes;
byte_count -= cur_size_in_bytes;
@ -1574,7 +1754,7 @@ static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev)
adev, &amdgpu_ttm_gtt_fops);
if (IS_ERR(ent))
return PTR_ERR(ent);
i_size_write(ent->d_inode, adev->mc.gtt_size);
i_size_write(ent->d_inode, adev->mc.gart_size);
adev->mman.gtt = ent;
#endif

18
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
View file

@ -34,6 +34,9 @@
#define AMDGPU_PL_FLAG_GWS (TTM_PL_FLAG_PRIV << 1)
#define AMDGPU_PL_FLAG_OA (TTM_PL_FLAG_PRIV << 2)
#define AMDGPU_GTT_MAX_TRANSFER_SIZE 512
#define AMDGPU_GTT_NUM_TRANSFER_WINDOWS 2
struct amdgpu_mman {
struct ttm_bo_global_ref bo_global_ref;
struct drm_global_reference mem_global_ref;
@ -49,6 +52,8 @@ struct amdgpu_mman {
/* buffer handling */
const struct amdgpu_buffer_funcs *buffer_funcs;
struct amdgpu_ring *buffer_funcs_ring;
struct mutex gtt_window_lock;
/* Scheduler entity for buffer moves */
struct amd_sched_entity entity;
};
@ -56,24 +61,25 @@ struct amdgpu_mman {
extern const struct ttm_mem_type_manager_func amdgpu_gtt_mgr_func;
extern const struct ttm_mem_type_manager_func amdgpu_vram_mgr_func;
bool amdgpu_gtt_mgr_is_allocated(struct ttm_mem_reg *mem);
int amdgpu_gtt_mgr_alloc(struct ttm_mem_type_manager *man,
struct ttm_buffer_object *tbo,
const struct ttm_place *place,
struct ttm_mem_reg *mem);
int amdgpu_copy_buffer(struct amdgpu_ring *ring,
uint64_t src_offset,
uint64_t dst_offset,
uint32_t byte_count,
int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset,
uint64_t dst_offset, uint32_t byte_count,
struct reservation_object *resv,
struct dma_fence **fence, bool direct_submit);
struct dma_fence **fence, bool direct_submit,
bool vm_needs_flush);
int amdgpu_fill_buffer(struct amdgpu_bo *bo,
uint32_t src_data,
uint64_t src_data,
struct reservation_object *resv,
struct dma_fence **fence);
int amdgpu_mmap(struct file *filp, struct vm_area_struct *vma);
bool amdgpu_ttm_is_bound(struct ttm_tt *ttm);
int amdgpu_ttm_bind(struct ttm_buffer_object *bo, struct ttm_mem_reg *bo_mem);
int amdgpu_ttm_recover_gart(struct amdgpu_device *adev);
#endif

18
drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
View file

@ -275,14 +275,10 @@ amdgpu_ucode_get_load_type(struct amdgpu_device *adev, int load_type)
else
return AMDGPU_FW_LOAD_PSP;
case CHIP_RAVEN:
#if 0
if (!load_type)
if (load_type != 2)
return AMDGPU_FW_LOAD_DIRECT;
else
return AMDGPU_FW_LOAD_PSP;
#else
return AMDGPU_FW_LOAD_DIRECT;
#endif
default:
DRM_ERROR("Unknow firmware load type\n");
}
@ -362,8 +358,6 @@ static int amdgpu_ucode_patch_jt(struct amdgpu_firmware_info *ucode,
(le32_to_cpu(header->jt_offset) * 4);
memcpy(dst_addr, src_addr, le32_to_cpu(header->jt_size) * 4);
ucode->ucode_size += le32_to_cpu(header->jt_size) * 4;
return 0;
}
@ -377,10 +371,15 @@ int amdgpu_ucode_init_bo(struct amdgpu_device *adev)
struct amdgpu_firmware_info *ucode = NULL;
const struct common_firmware_header *header = NULL;
if (!adev->firmware.fw_size) {
dev_warn(adev->dev, "No ip firmware need to load\n");
return 0;
}
err = amdgpu_bo_create(adev, adev->firmware.fw_size, PAGE_SIZE, true,
amdgpu_sriov_vf(adev) ? AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT,
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, bo);
NULL, NULL, 0, bo);
if (err) {
dev_err(adev->dev, "(%d) Firmware buffer allocate failed\n", err);
goto failed;
@ -459,6 +458,9 @@ int amdgpu_ucode_fini_bo(struct amdgpu_device *adev)
int i;
struct amdgpu_firmware_info *ucode = NULL;
if (!adev->firmware.fw_size)
return 0;
for (i = 0; i < adev->firmware.max_ucodes; i++) {
ucode = &adev->firmware.ucode[i];
if (ucode->fw) {

4
drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
View file

@ -1051,7 +1051,7 @@ int amdgpu_uvd_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &bo);
NULL, NULL, 0, &bo);
if (r)
return r;
@ -1101,7 +1101,7 @@ int amdgpu_uvd_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &bo);
NULL, NULL, 0, &bo);
if (r)
return r;

4
drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
View file

@ -937,9 +937,9 @@ int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring)
unsigned i;
int r, timeout = adev->usec_timeout;
/* workaround VCE ring test slow issue for sriov*/
/* skip ring test for sriov*/
if (amdgpu_sriov_vf(adev))
timeout *= 10;
return 0;
r = amdgpu_ring_alloc(ring, 16);
if (r) {

18
drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c
View file

@ -209,9 +209,9 @@ static void amdgpu_vcn_idle_work_handler(struct work_struct *work)
if (fences == 0) {
if (adev->pm.dpm_enabled) {
/* might be used when with pg/cg
amdgpu_dpm_enable_uvd(adev, false);
} else {
amdgpu_asic_set_uvd_clocks(adev, 0, 0);
*/
}
} else {
schedule_delayed_work(&adev->vcn.idle_work, VCN_IDLE_TIMEOUT);
@ -223,12 +223,10 @@ void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring)
struct amdgpu_device *adev = ring->adev;
bool set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
if (set_clocks) {
if (adev->pm.dpm_enabled) {
amdgpu_dpm_enable_uvd(adev, true);
} else {
amdgpu_asic_set_uvd_clocks(adev, 53300, 40000);
}
if (set_clocks && adev->pm.dpm_enabled) {
/* might be used when with pg/cg
amdgpu_dpm_enable_uvd(adev, true);
*/
}
}
@ -361,7 +359,7 @@ static int amdgpu_vcn_dec_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &bo);
NULL, NULL, 0, &bo);
if (r)
return r;
@ -413,7 +411,7 @@ static int amdgpu_vcn_dec_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL, &bo);
NULL, NULL, 0, &bo);
if (r)
return r;

85
drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.c Normal file
View file

@ -0,0 +1,85 @@
/*
* Copyright 2017 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_vf_error.h"
#include "mxgpu_ai.h"
#define AMDGPU_VF_ERROR_ENTRY_SIZE 16
/* struct error_entry - amdgpu VF error information. */
struct amdgpu_vf_error_buffer {
int read_count;
int write_count;
uint16_t code[AMDGPU_VF_ERROR_ENTRY_SIZE];
uint16_t flags[AMDGPU_VF_ERROR_ENTRY_SIZE];
uint64_t data[AMDGPU_VF_ERROR_ENTRY_SIZE];
};
struct amdgpu_vf_error_buffer admgpu_vf_errors;
void amdgpu_vf_error_put(uint16_t sub_error_code, uint16_t error_flags, uint64_t error_data)
{
int index;
uint16_t error_code = AMDGIM_ERROR_CODE(AMDGIM_ERROR_CATEGORY_VF, sub_error_code);
index = admgpu_vf_errors.write_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
admgpu_vf_errors.code [index] = error_code;
admgpu_vf_errors.flags [index] = error_flags;
admgpu_vf_errors.data [index] = error_data;
admgpu_vf_errors.write_count ++;
}
void amdgpu_vf_error_trans_all(struct amdgpu_device *adev)
{
/* u32 pf2vf_flags = 0; */
u32 data1, data2, data3;
int index;
if ((NULL == adev) || (!amdgpu_sriov_vf(adev)) || (!adev->virt.ops) || (!adev->virt.ops->trans_msg)) {
return;
}
/*
TODO: Enable these code when pv2vf_info is merged
AMDGPU_FW_VRAM_PF2VF_READ (adev, feature_flags, &pf2vf_flags);
if (!(pf2vf_flags & AMDGIM_FEATURE_ERROR_LOG_COLLECT)) {
return;
}
*/
/* The errors are overlay of array, correct read_count as full. */
if (admgpu_vf_errors.write_count - admgpu_vf_errors.read_count > AMDGPU_VF_ERROR_ENTRY_SIZE) {
admgpu_vf_errors.read_count = admgpu_vf_errors.write_count - AMDGPU_VF_ERROR_ENTRY_SIZE;
}
while (admgpu_vf_errors.read_count < admgpu_vf_errors.write_count) {
index =admgpu_vf_errors.read_count % AMDGPU_VF_ERROR_ENTRY_SIZE;
data1 = AMDGIM_ERROR_CODE_FLAGS_TO_MAILBOX (admgpu_vf_errors.code[index], admgpu_vf_errors.flags[index]);
data2 = admgpu_vf_errors.data[index] & 0xFFFFFFFF;
data3 = (admgpu_vf_errors.data[index] >> 32) & 0xFFFFFFFF;
adev->virt.ops->trans_msg(adev, IDH_LOG_VF_ERROR, data1, data2, data3);
admgpu_vf_errors.read_count ++;
}
}

62
drivers/gpu/drm/amd/amdgpu/amdgpu_vf_error.h Normal file
View file

@ -0,0 +1,62 @@
/*
* Copyright 2017 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 __VF_ERROR_H__
#define __VF_ERROR_H__
#define AMDGIM_ERROR_CODE_FLAGS_TO_MAILBOX(c,f) (((c & 0xFFFF) << 16) | (f & 0xFFFF))
#define AMDGIM_ERROR_CODE(t,c) (((t&0xF)<<12)|(c&0xFFF))
/* Please keep enum same as AMD GIM driver */
enum AMDGIM_ERROR_VF {
AMDGIM_ERROR_VF_ATOMBIOS_INIT_FAIL = 0,
AMDGIM_ERROR_VF_NO_VBIOS,
AMDGIM_ERROR_VF_GPU_POST_ERROR,
AMDGIM_ERROR_VF_ATOMBIOS_GET_CLOCK_FAIL,
AMDGIM_ERROR_VF_FENCE_INIT_FAIL,
AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL,
AMDGIM_ERROR_VF_IB_INIT_FAIL,
AMDGIM_ERROR_VF_AMDGPU_LATE_INIT_FAIL,
AMDGIM_ERROR_VF_ASIC_RESUME_FAIL,
AMDGIM_ERROR_VF_GPU_RESET_FAIL,
AMDGIM_ERROR_VF_TEST,
AMDGIM_ERROR_VF_MAX
};
enum AMDGIM_ERROR_CATEGORY {
AMDGIM_ERROR_CATEGORY_NON_USED = 0,
AMDGIM_ERROR_CATEGORY_GIM,
AMDGIM_ERROR_CATEGORY_PF,
AMDGIM_ERROR_CATEGORY_VF,
AMDGIM_ERROR_CATEGORY_VBIOS,
AMDGIM_ERROR_CATEGORY_MONITOR,
AMDGIM_ERROR_CATEGORY_MAX
};
void amdgpu_vf_error_put(uint16_t sub_error_code, uint16_t error_flags, uint64_t error_data);
void amdgpu_vf_error_trans_all (struct amdgpu_device *adev);
#endif /* __VF_ERROR_H__ */

1
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
View file

@ -43,6 +43,7 @@ struct amdgpu_virt_ops {
int (*req_full_gpu)(struct amdgpu_device *adev, bool init);
int (*rel_full_gpu)(struct amdgpu_device *adev, bool init);
int (*reset_gpu)(struct amdgpu_device *adev);
void (*trans_msg)(struct amdgpu_device *adev, u32 req, u32 data1, u32 data2, u32 data3);
};
/* GPU virtualization */

316
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
View file

@ -77,8 +77,6 @@ struct amdgpu_pte_update_params {
void (*func)(struct amdgpu_pte_update_params *params, uint64_t pe,
uint64_t addr, unsigned count, uint32_t incr,
uint64_t flags);
/* indicate update pt or its shadow */
bool shadow;
/* The next two are used during VM update by CPU
* DMA addresses to use for mapping
* Kernel pointer of PD/PT BO that needs to be updated
@ -161,11 +159,17 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
*/
static int amdgpu_vm_validate_level(struct amdgpu_vm_pt *parent,
int (*validate)(void *, struct amdgpu_bo *),
void *param)
void *param, bool use_cpu_for_update)
{
unsigned i;
int r;
if (use_cpu_for_update) {
r = amdgpu_bo_kmap(parent->bo, NULL);
if (r)
return r;
}
if (!parent->entries)
return 0;
@ -183,7 +187,8 @@ static int amdgpu_vm_validate_level(struct amdgpu_vm_pt *parent,
* Recurse into the sub directory. This is harmless because we
* have only a maximum of 5 layers.
*/
r = amdgpu_vm_validate_level(entry, validate, param);
r = amdgpu_vm_validate_level(entry, validate, param,
use_cpu_for_update);
if (r)
return r;
}
@ -214,7 +219,8 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
if (num_evictions == vm->last_eviction_counter)
return 0;
return amdgpu_vm_validate_level(&vm->root, validate, param);
return amdgpu_vm_validate_level(&vm->root, validate, param,
vm->use_cpu_for_update);
}
/**
@ -282,6 +288,7 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
unsigned pt_idx, from, to;
int r;
u64 flags;
uint64_t init_value = 0;
if (!parent->entries) {
unsigned num_entries = amdgpu_vm_num_entries(adev, level);
@ -315,6 +322,12 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
flags |= (AMDGPU_GEM_CREATE_NO_CPU_ACCESS |
AMDGPU_GEM_CREATE_SHADOW);
if (vm->pte_support_ats) {
init_value = AMDGPU_PTE_SYSTEM;
if (level != adev->vm_manager.num_level - 1)
init_value |= AMDGPU_PDE_PTE;
}
/* walk over the address space and allocate the page tables */
for (pt_idx = from; pt_idx <= to; ++pt_idx) {
struct reservation_object *resv = vm->root.bo->tbo.resv;
@ -327,10 +340,18 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
AMDGPU_GPU_PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
flags,
NULL, resv, &pt);
NULL, resv, init_value, &pt);
if (r)
return r;
if (vm->use_cpu_for_update) {
r = amdgpu_bo_kmap(pt, NULL);
if (r) {
amdgpu_bo_unref(&pt);
return r;
}
}
/* Keep a reference to the root directory to avoid
* freeing them up in the wrong order.
*/
@ -338,6 +359,7 @@ static int amdgpu_vm_alloc_levels(struct amdgpu_device *adev,
entry->bo = pt;
entry->addr = 0;
entry->huge_page = false;
}
if (level < adev->vm_manager.num_level) {
@ -424,7 +446,7 @@ static int amdgpu_vm_grab_reserved_vmid_locked(struct amdgpu_vm *vm,
struct dma_fence *updates = sync->last_vm_update;
int r = 0;
struct dma_fence *flushed, *tmp;
bool needs_flush = false;
bool needs_flush = vm->use_cpu_for_update;
flushed = id->flushed_updates;
if ((amdgpu_vm_had_gpu_reset(adev, id)) ||
@ -545,11 +567,11 @@ int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
}
kfree(fences);
job->vm_needs_flush = false;
job->vm_needs_flush = vm->use_cpu_for_update;
/* Check if we can use a VMID already assigned to this VM */
list_for_each_entry_reverse(id, &id_mgr->ids_lru, list) {
struct dma_fence *flushed;
bool needs_flush = false;
bool needs_flush = vm->use_cpu_for_update;
/* Check all the prerequisites to using this VMID */
if (amdgpu_vm_had_gpu_reset(adev, id))
@ -745,7 +767,7 @@ static bool amdgpu_vm_is_large_bar(struct amdgpu_device *adev)
*
* Emit a VM flush when it is necessary.
*/
int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job)
int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync)
{
struct amdgpu_device *adev = ring->adev;
unsigned vmhub = ring->funcs->vmhub;
@ -767,12 +789,15 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job)
vm_flush_needed = true;
}
if (!vm_flush_needed && !gds_switch_needed)
if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync)
return 0;
if (ring->funcs->init_cond_exec)
patch_offset = amdgpu_ring_init_cond_exec(ring);
if (need_pipe_sync)
amdgpu_ring_emit_pipeline_sync(ring);
if (ring->funcs->emit_vm_flush && vm_flush_needed) {
struct dma_fence *fence;
@ -981,6 +1006,8 @@ static void amdgpu_vm_cpu_set_ptes(struct amdgpu_pte_update_params *params,
unsigned int i;
uint64_t value;
trace_amdgpu_vm_set_ptes(pe, addr, count, incr, flags);
for (i = 0; i < count; i++) {
value = params->pages_addr ?
amdgpu_vm_map_gart(params->pages_addr, addr) :
@ -989,19 +1016,16 @@ static void amdgpu_vm_cpu_set_ptes(struct amdgpu_pte_update_params *params,
i, value, flags);
addr += incr;
}
/* Flush HDP */
mb();
amdgpu_gart_flush_gpu_tlb(params->adev, 0);
}
static int amdgpu_vm_bo_wait(struct amdgpu_device *adev, struct amdgpu_bo *bo)
static int amdgpu_vm_wait_pd(struct amdgpu_device *adev, struct amdgpu_vm *vm,
void *owner)
{
struct amdgpu_sync sync;
int r;
amdgpu_sync_create(&sync);
amdgpu_sync_resv(adev, &sync, bo->tbo.resv, AMDGPU_FENCE_OWNER_VM);
amdgpu_sync_resv(adev, &sync, vm->root.bo->tbo.resv, owner);
r = amdgpu_sync_wait(&sync, true);
amdgpu_sync_free(&sync);
@ -1042,16 +1066,12 @@ static int amdgpu_vm_update_level(struct amdgpu_device *adev,
params.adev = adev;
shadow = parent->bo->shadow;
WARN_ON(vm->use_cpu_for_update && shadow);
if (vm->use_cpu_for_update && !shadow) {
r = amdgpu_bo_kmap(parent->bo, (void **)&pd_addr);
if (r)
if (vm->use_cpu_for_update) {
pd_addr = (unsigned long)amdgpu_bo_kptr(parent->bo);
r = amdgpu_vm_wait_pd(adev, vm, AMDGPU_FENCE_OWNER_VM);
if (unlikely(r))
return r;
r = amdgpu_vm_bo_wait(adev, parent->bo);
if (unlikely(r)) {
amdgpu_bo_kunmap(parent->bo);
return r;
}
params.func = amdgpu_vm_cpu_set_ptes;
} else {
if (shadow) {
@ -1105,7 +1125,8 @@ static int amdgpu_vm_update_level(struct amdgpu_device *adev,
pt = amdgpu_bo_gpu_offset(bo);
pt = amdgpu_gart_get_vm_pde(adev, pt);
if (parent->entries[pt_idx].addr == pt)
if (parent->entries[pt_idx].addr == pt ||
parent->entries[pt_idx].huge_page)
continue;
parent->entries[pt_idx].addr = pt;
@ -1146,28 +1167,29 @@ static int amdgpu_vm_update_level(struct amdgpu_device *adev,
count, incr, AMDGPU_PTE_VALID);
}
if (params.func == amdgpu_vm_cpu_set_ptes)
amdgpu_bo_kunmap(parent->bo);
else if (params.ib->length_dw == 0) {
amdgpu_job_free(job);
} else {
amdgpu_ring_pad_ib(ring, params.ib);
amdgpu_sync_resv(adev, &job->sync, parent->bo->tbo.resv,
AMDGPU_FENCE_OWNER_VM);
if (shadow)
amdgpu_sync_resv(adev, &job->sync, shadow->tbo.resv,
if (!vm->use_cpu_for_update) {
if (params.ib->length_dw == 0) {
amdgpu_job_free(job);
} else {
amdgpu_ring_pad_ib(ring, params.ib);
amdgpu_sync_resv(adev, &job->sync, parent->bo->tbo.resv,
AMDGPU_FENCE_OWNER_VM);
if (shadow)
amdgpu_sync_resv(adev, &job->sync,
shadow->tbo.resv,
AMDGPU_FENCE_OWNER_VM);
WARN_ON(params.ib->length_dw > ndw);
r = amdgpu_job_submit(job, ring, &vm->entity,
AMDGPU_FENCE_OWNER_VM, &fence);
if (r)
goto error_free;
WARN_ON(params.ib->length_dw > ndw);
r = amdgpu_job_submit(job, ring, &vm->entity,
AMDGPU_FENCE_OWNER_VM, &fence);
if (r)
goto error_free;
amdgpu_bo_fence(parent->bo, fence, true);
dma_fence_put(vm->last_dir_update);
vm->last_dir_update = dma_fence_get(fence);
dma_fence_put(fence);
amdgpu_bo_fence(parent->bo, fence, true);
dma_fence_put(vm->last_dir_update);
vm->last_dir_update = dma_fence_get(fence);
dma_fence_put(fence);
}
}
/*
* Recurse into the subdirectories. This recursion is harmless because
@ -1235,33 +1257,105 @@ int amdgpu_vm_update_directories(struct amdgpu_device *adev,
if (r)
amdgpu_vm_invalidate_level(&vm->root);
if (vm->use_cpu_for_update) {
/* Flush HDP */
mb();
amdgpu_gart_flush_gpu_tlb(adev, 0);
}
return r;
}
/**
* amdgpu_vm_find_pt - find the page table for an address
* amdgpu_vm_find_entry - find the entry for an address
*
* @p: see amdgpu_pte_update_params definition
* @addr: virtual address in question
* @entry: resulting entry or NULL
* @parent: parent entry
*
* Find the page table BO for a virtual address, return NULL when none found.
* Find the vm_pt entry and it's parent for the given address.
*/
static struct amdgpu_bo *amdgpu_vm_get_pt(struct amdgpu_pte_update_params *p,
uint64_t addr)
void amdgpu_vm_get_entry(struct amdgpu_pte_update_params *p, uint64_t addr,
struct amdgpu_vm_pt **entry,
struct amdgpu_vm_pt **parent)
{
struct amdgpu_vm_pt *entry = &p->vm->root;
unsigned idx, level = p->adev->vm_manager.num_level;
while (entry->entries) {
*parent = NULL;
*entry = &p->vm->root;
while ((*entry)->entries) {
idx = addr >> (p->adev->vm_manager.block_size * level--);
idx %= amdgpu_bo_size(entry->bo) / 8;
entry = &entry->entries[idx];
idx %= amdgpu_bo_size((*entry)->bo) / 8;
*parent = *entry;
*entry = &(*entry)->entries[idx];
}
if (level)
return NULL;
*entry = NULL;
}
return entry->bo;
/**
* amdgpu_vm_handle_huge_pages - handle updating the PD with huge pages
*
* @p: see amdgpu_pte_update_params definition
* @entry: vm_pt entry to check
* @parent: parent entry
* @nptes: number of PTEs updated with this operation
* @dst: destination address where the PTEs should point to
* @flags: access flags fro the PTEs
*
* Check if we can update the PD with a huge page.
*/
static int amdgpu_vm_handle_huge_pages(struct amdgpu_pte_update_params *p,
struct amdgpu_vm_pt *entry,
struct amdgpu_vm_pt *parent,
unsigned nptes, uint64_t dst,
uint64_t flags)
{
bool use_cpu_update = (p->func == amdgpu_vm_cpu_set_ptes);
uint64_t pd_addr, pde;
int r;
/* In the case of a mixed PT the PDE must point to it*/
if (p->adev->asic_type < CHIP_VEGA10 ||
nptes != AMDGPU_VM_PTE_COUNT(p->adev) ||
p->func == amdgpu_vm_do_copy_ptes ||
!(flags & AMDGPU_PTE_VALID)) {
dst = amdgpu_bo_gpu_offset(entry->bo);
dst = amdgpu_gart_get_vm_pde(p->adev, dst);
flags = AMDGPU_PTE_VALID;
} else {
flags |= AMDGPU_PDE_PTE;
}
if (entry->addr == dst &&
entry->huge_page == !!(flags & AMDGPU_PDE_PTE))
return 0;
entry->addr = dst;
entry->huge_page = !!(flags & AMDGPU_PDE_PTE);
if (use_cpu_update) {
r = amdgpu_bo_kmap(parent->bo, (void *)&pd_addr);
if (r)
return r;
pde = pd_addr + (entry - parent->entries) * 8;
amdgpu_vm_cpu_set_ptes(p, pde, dst, 1, 0, flags);
} else {
if (parent->bo->shadow) {
pd_addr = amdgpu_bo_gpu_offset(parent->bo->shadow);
pde = pd_addr + (entry - parent->entries) * 8;
amdgpu_vm_do_set_ptes(p, pde, dst, 1, 0, flags);
}
pd_addr = amdgpu_bo_gpu_offset(parent->bo);
pde = pd_addr + (entry - parent->entries) * 8;
amdgpu_vm_do_set_ptes(p, pde, dst, 1, 0, flags);
}
return 0;
}
/**
@ -1287,49 +1381,47 @@ static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params,
uint64_t addr, pe_start;
struct amdgpu_bo *pt;
unsigned nptes;
int r;
bool use_cpu_update = (params->func == amdgpu_vm_cpu_set_ptes);
int r;
/* walk over the address space and update the page tables */
for (addr = start; addr < end; addr += nptes) {
pt = amdgpu_vm_get_pt(params, addr);
if (!pt) {
pr_err("PT not found, aborting update_ptes\n");
return -EINVAL;
}
for (addr = start; addr < end; addr += nptes,
dst += nptes * AMDGPU_GPU_PAGE_SIZE) {
struct amdgpu_vm_pt *entry, *parent;
if (params->shadow) {
if (WARN_ONCE(use_cpu_update,
"CPU VM update doesn't suuport shadow pages"))
return 0;
if (!pt->shadow)
return 0;
pt = pt->shadow;
}
amdgpu_vm_get_entry(params, addr, &entry, &parent);
if (!entry)
return -ENOENT;
if ((addr & ~mask) == (end & ~mask))
nptes = end - addr;
else
nptes = AMDGPU_VM_PTE_COUNT(adev) - (addr & mask);
r = amdgpu_vm_handle_huge_pages(params, entry, parent,
nptes, dst, flags);
if (r)
return r;
if (entry->huge_page)
continue;
pt = entry->bo;
if (use_cpu_update) {
r = amdgpu_bo_kmap(pt, (void *)&pe_start);
if (r)
return r;
} else
pe_start = (unsigned long)amdgpu_bo_kptr(pt);
} else {
if (pt->shadow) {
pe_start = amdgpu_bo_gpu_offset(pt->shadow);
pe_start += (addr & mask) * 8;
params->func(params, pe_start, dst, nptes,
AMDGPU_GPU_PAGE_SIZE, flags);
}
pe_start = amdgpu_bo_gpu_offset(pt);
}
pe_start += (addr & mask) * 8;
params->func(params, pe_start, dst, nptes,
AMDGPU_GPU_PAGE_SIZE, flags);
dst += nptes * AMDGPU_GPU_PAGE_SIZE;
if (use_cpu_update)
amdgpu_bo_kunmap(pt);
}
return 0;
@ -1372,8 +1464,9 @@ static int amdgpu_vm_frag_ptes(struct amdgpu_pte_update_params *params,
*/
/* SI and newer are optimized for 64KB */
uint64_t frag_flags = AMDGPU_PTE_FRAG(AMDGPU_LOG2_PAGES_PER_FRAG);
uint64_t frag_align = 1 << AMDGPU_LOG2_PAGES_PER_FRAG;
unsigned pages_per_frag = AMDGPU_LOG2_PAGES_PER_FRAG(params->adev);
uint64_t frag_flags = AMDGPU_PTE_FRAG(pages_per_frag);
uint64_t frag_align = 1 << pages_per_frag;
uint64_t frag_start = ALIGN(start, frag_align);
uint64_t frag_end = end & ~(frag_align - 1);
@ -1445,6 +1538,10 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
params.vm = vm;
params.src = src;
/* sync to everything on unmapping */
if (!(flags & AMDGPU_PTE_VALID))
owner = AMDGPU_FENCE_OWNER_UNDEFINED;
if (vm->use_cpu_for_update) {
/* params.src is used as flag to indicate system Memory */
if (pages_addr)
@ -1453,23 +1550,18 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
/* Wait for PT BOs to be free. PTs share the same resv. object
* as the root PD BO
*/
r = amdgpu_vm_bo_wait(adev, vm->root.bo);
r = amdgpu_vm_wait_pd(adev, vm, owner);
if (unlikely(r))
return r;
params.func = amdgpu_vm_cpu_set_ptes;
params.pages_addr = pages_addr;
params.shadow = false;
return amdgpu_vm_frag_ptes(&params, start, last + 1,
addr, flags);
}
ring = container_of(vm->entity.sched, struct amdgpu_ring, sched);
/* sync to everything on unmapping */
if (!(flags & AMDGPU_PTE_VALID))
owner = AMDGPU_FENCE_OWNER_UNDEFINED;
nptes = last - start + 1;
/*
@ -1481,6 +1573,9 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
/* padding, etc. */
ndw = 64;
/* one PDE write for each huge page */
ndw += ((nptes >> adev->vm_manager.block_size) + 1) * 6;
if (src) {
/* only copy commands needed */
ndw += ncmds * 7;
@ -1542,11 +1637,6 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
if (r)
goto error_free;
params.shadow = true;
r = amdgpu_vm_frag_ptes(&params, start, last + 1, addr, flags);
if (r)
goto error_free;
params.shadow = false;
r = amdgpu_vm_frag_ptes(&params, start, last + 1, addr, flags);
if (r)
goto error_free;
@ -1565,6 +1655,7 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
error_free:
amdgpu_job_free(job);
amdgpu_vm_invalidate_level(&vm->root);
return r;
}
@ -1752,6 +1843,12 @@ int amdgpu_vm_bo_update(struct amdgpu_device *adev,
list_add(&bo_va->vm_status, &vm->cleared);
spin_unlock(&vm->status_lock);
if (vm->use_cpu_for_update) {
/* Flush HDP */
mb();
amdgpu_gart_flush_gpu_tlb(adev, 0);
}
return 0;
}
@ -1905,15 +2002,19 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
struct amdgpu_bo_va_mapping *mapping;
struct dma_fence *f = NULL;
int r;
uint64_t init_pte_value = 0;
while (!list_empty(&vm->freed)) {
mapping = list_first_entry(&vm->freed,
struct amdgpu_bo_va_mapping, list);
list_del(&mapping->list);
if (vm->pte_support_ats)
init_pte_value = AMDGPU_PTE_SYSTEM;
r = amdgpu_vm_bo_update_mapping(adev, NULL, 0, NULL, vm,
mapping->start, mapping->last,
0, 0, &f);
init_pte_value, 0, &f);
amdgpu_vm_free_mapping(adev, vm, mapping, f);
if (r) {
dma_fence_put(f);
@ -2404,6 +2505,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
struct amd_sched_rq *rq;
int r, i;
u64 flags;
uint64_t init_pde_value = 0;
vm->va = RB_ROOT;
vm->client_id = atomic64_inc_return(&adev->vm_manager.client_counter);
@ -2425,10 +2527,17 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
if (r)
return r;
if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE)
vm->pte_support_ats = false;
if (vm_context == AMDGPU_VM_CONTEXT_COMPUTE) {
vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
AMDGPU_VM_USE_CPU_FOR_COMPUTE);
else
if (adev->asic_type == CHIP_RAVEN) {
vm->pte_support_ats = true;
init_pde_value = AMDGPU_PTE_SYSTEM | AMDGPU_PDE_PTE;
}
} else
vm->use_cpu_for_update = !!(adev->vm_manager.vm_update_mode &
AMDGPU_VM_USE_CPU_FOR_GFX);
DRM_DEBUG_DRIVER("VM update mode is %s\n",
@ -2448,7 +2557,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
r = amdgpu_bo_create(adev, amdgpu_vm_bo_size(adev, 0), align, true,
AMDGPU_GEM_DOMAIN_VRAM,
flags,
NULL, NULL, &vm->root.bo);
NULL, NULL, init_pde_value, &vm->root.bo);
if (r)
goto error_free_sched_entity;
@ -2457,6 +2566,13 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
goto error_free_root;
vm->last_eviction_counter = atomic64_read(&adev->num_evictions);
if (vm->use_cpu_for_update) {
r = amdgpu_bo_kmap(vm->root.bo, NULL);
if (r)
goto error_free_root;
}
amdgpu_bo_unreserve(vm->root.bo);
return 0;

13
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
View file

@ -51,7 +51,9 @@ struct amdgpu_bo_list_entry;
#define AMDGPU_VM_PTB_ALIGN_SIZE 32768
/* LOG2 number of continuous pages for the fragment field */
#define AMDGPU_LOG2_PAGES_PER_FRAG 4
#define AMDGPU_LOG2_PAGES_PER_FRAG(adev) \
((adev)->asic_type < CHIP_VEGA10 ? 4 : \
(adev)->vm_manager.block_size)
#define AMDGPU_PTE_VALID (1ULL << 0)
#define AMDGPU_PTE_SYSTEM (1ULL << 1)
@ -68,6 +70,9 @@ struct amdgpu_bo_list_entry;
/* TILED for VEGA10, reserved for older ASICs */
#define AMDGPU_PTE_PRT (1ULL << 51)
/* PDE is handled as PTE for VEGA10 */
#define AMDGPU_PDE_PTE (1ULL << 54)
/* VEGA10 only */
#define AMDGPU_PTE_MTYPE(a) ((uint64_t)a << 57)
#define AMDGPU_PTE_MTYPE_MASK AMDGPU_PTE_MTYPE(3ULL)
@ -98,6 +103,7 @@ struct amdgpu_bo_list_entry;
struct amdgpu_vm_pt {
struct amdgpu_bo *bo;
uint64_t addr;
bool huge_page;
/* array of page tables, one for each directory entry */
struct amdgpu_vm_pt *entries;
@ -140,6 +146,9 @@ struct amdgpu_vm {
/* Flag to indicate if VM tables are updated by CPU or GPU (SDMA) */
bool use_cpu_for_update;
/* Flag to indicate ATS support from PTE for GFX9 */
bool pte_support_ats;
};
struct amdgpu_vm_id {
@ -222,7 +231,7 @@ int amdgpu_vm_alloc_pts(struct amdgpu_device *adev,
int amdgpu_vm_grab_id(struct amdgpu_vm *vm, struct amdgpu_ring *ring,
struct amdgpu_sync *sync, struct dma_fence *fence,
struct amdgpu_job *job);
int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job);
int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync);
void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vmhub,
unsigned vmid);
void amdgpu_vm_reset_all_ids(struct amdgpu_device *adev);

9
drivers/gpu/drm/amd/amdgpu/cik.c
View file

@ -1824,21 +1824,14 @@ static int cik_common_suspend(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
amdgpu_amdkfd_suspend(adev);
return cik_common_hw_fini(adev);
}
static int cik_common_resume(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
r = cik_common_hw_init(adev);
if (r)
return r;
return amdgpu_amdkfd_resume(adev);
return cik_common_hw_init(adev);
}
static bool cik_common_is_idle(void *handle)

60
drivers/gpu/drm/amd/amdgpu/cik_sdma.c
View file

@ -341,6 +341,63 @@ static void cik_sdma_rlc_stop(struct amdgpu_device *adev)
/* XXX todo */
}
/**
* cik_ctx_switch_enable - stop the async dma engines context switch
*
* @adev: amdgpu_device pointer
* @enable: enable/disable the DMA MEs context switch.
*
* Halt or unhalt the async dma engines context switch (VI).
*/
static void cik_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
{
u32 f32_cntl, phase_quantum = 0;
int i;
if (amdgpu_sdma_phase_quantum) {
unsigned value = amdgpu_sdma_phase_quantum;
unsigned unit = 0;
while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
value = (value + 1) >> 1;
unit++;
}
if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
WARN_ONCE(1,
"clamping sdma_phase_quantum to %uK clock cycles\n",
value << unit);
}
phase_quantum =
value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
unit << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
}
for (i = 0; i < adev->sdma.num_instances; i++) {
f32_cntl = RREG32(mmSDMA0_CNTL + sdma_offsets[i]);
if (enable) {
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
AUTO_CTXSW_ENABLE, 1);
if (amdgpu_sdma_phase_quantum) {
WREG32(mmSDMA0_PHASE0_QUANTUM + sdma_offsets[i],
phase_quantum);
WREG32(mmSDMA0_PHASE1_QUANTUM + sdma_offsets[i],
phase_quantum);
}
} else {
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
AUTO_CTXSW_ENABLE, 0);
}
WREG32(mmSDMA0_CNTL + sdma_offsets[i], f32_cntl);
}
}
/**
* cik_sdma_enable - stop the async dma engines
*
@ -537,6 +594,8 @@ static int cik_sdma_start(struct amdgpu_device *adev)
/* halt the engine before programing */
cik_sdma_enable(adev, false);
/* enable sdma ring preemption */
cik_ctx_switch_enable(adev, true);
/* start the gfx rings and rlc compute queues */
r = cik_sdma_gfx_resume(adev);
@ -984,6 +1043,7 @@ static int cik_sdma_hw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
cik_ctx_switch_enable(adev, false);
cik_sdma_enable(adev, false);
return 0;

41
drivers/gpu/drm/amd/amdgpu/clearstate_gfx9.h
View file

@ -1,24 +1,25 @@
/*
***************************************************************************************************
*
* Trade secret of Advanced Micro Devices, Inc.
* Copyright (c) 2010 Advanced Micro Devices, Inc. (unpublished)
*
* All rights reserved. This notice is intended as a precaution against inadvertent publication and
* does not imply publication or any waiver of confidentiality. The year included in the foregoing
* notice is the year of creation of the work.
*
***************************************************************************************************
*/
/**
***************************************************************************************************
* @brief gfx9 Clearstate Definitions
***************************************************************************************************
*
* Do not edit! This is a machine-generated file!
*
*/
* Copyright 2017 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 gfx9_SECT_CONTEXT_def_1[] =
{

139
drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
View file

@ -484,134 +484,6 @@ static bool dce_v10_0_is_display_hung(struct amdgpu_device *adev)
return true;
}
static void dce_v10_0_stop_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 crtc_enabled, tmp;
int i;
save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
/* disable VGA render */
tmp = RREG32(mmVGA_RENDER_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
WREG32(mmVGA_RENDER_CONTROL, tmp);
/* blank the display controllers */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
CRTC_CONTROL, CRTC_MASTER_EN);
if (crtc_enabled) {
#if 0
u32 frame_count;
int j;
save->crtc_enabled[i] = true;
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
amdgpu_display_vblank_wait(adev, i);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
}
/* wait for the next frame */
frame_count = amdgpu_display_vblank_get_counter(adev, i);
for (j = 0; j < adev->usec_timeout; j++) {
if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
break;
udelay(1);
}
tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK) == 0) {
tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 1);
WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
}
tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK) == 0) {
tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 1);
WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
}
#else
/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
save->crtc_enabled[i] = false;
/* ***** */
#endif
} else {
save->crtc_enabled[i] = false;
}
}
}
static void dce_v10_0_resume_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 tmp, frame_count;
int i, j;
/* update crtc base addresses */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
upper_32_bits(adev->mc.vram_start));
WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
upper_32_bits(adev->mc.vram_start));
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
(u32)adev->mc.vram_start);
WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i],
(u32)adev->mc.vram_start);
if (save->crtc_enabled[i]) {
tmp = RREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE) != 0) {
tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_MODE, MASTER_UPDATE_MODE, 0);
WREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i], tmp);
}
tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK)) {
tmp = REG_SET_FIELD(tmp, GRPH_UPDATE, GRPH_UPDATE_LOCK, 0);
WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
}
tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK)) {
tmp = REG_SET_FIELD(tmp, MASTER_UPDATE_LOCK, MASTER_UPDATE_LOCK, 0);
WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
}
for (j = 0; j < adev->usec_timeout; j++) {
tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, GRPH_UPDATE, GRPH_SURFACE_UPDATE_PENDING) == 0)
break;
udelay(1);
}
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
/* wait for the next frame */
frame_count = amdgpu_display_vblank_get_counter(adev, i);
for (j = 0; j < adev->usec_timeout; j++) {
if (amdgpu_display_vblank_get_counter(adev, i) != frame_count)
break;
udelay(1);
}
}
}
WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
WREG32(mmVGA_MEMORY_BASE_ADDRESS, lower_32_bits(adev->mc.vram_start));
/* Unlock vga access */
WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
mdelay(1);
WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
}
static void dce_v10_0_set_vga_render_state(struct amdgpu_device *adev,
bool render)
{
@ -2559,7 +2431,7 @@ static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
aobj = gem_to_amdgpu_bo(obj);
ret = amdgpu_bo_reserve(aobj, false);
if (ret != 0) {
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2567,7 +2439,7 @@ static int dce_v10_0_crtc_cursor_set2(struct drm_crtc *crtc,
amdgpu_bo_unreserve(aobj);
if (ret) {
DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2601,7 +2473,7 @@ unpin:
amdgpu_bo_unpin(aobj);
amdgpu_bo_unreserve(aobj);
}
drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
}
amdgpu_crtc->cursor_bo = obj;
@ -3012,6 +2884,8 @@ static int dce_v10_0_hw_init(void *handle)
dce_v10_0_init_golden_registers(adev);
/* disable vga render */
dce_v10_0_set_vga_render_state(adev, false);
/* init dig PHYs, disp eng pll */
amdgpu_atombios_encoder_init_dig(adev);
amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
@ -3724,7 +3598,6 @@ static void dce_v10_0_encoder_add(struct amdgpu_device *adev,
}
static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
.set_vga_render_state = &dce_v10_0_set_vga_render_state,
.bandwidth_update = &dce_v10_0_bandwidth_update,
.vblank_get_counter = &dce_v10_0_vblank_get_counter,
.vblank_wait = &dce_v10_0_vblank_wait,
@ -3737,8 +3610,6 @@ static const struct amdgpu_display_funcs dce_v10_0_display_funcs = {
.page_flip_get_scanoutpos = &dce_v10_0_crtc_get_scanoutpos,
.add_encoder = &dce_v10_0_encoder_add,
.add_connector = &amdgpu_connector_add,
.stop_mc_access = &dce_v10_0_stop_mc_access,
.resume_mc_access = &dce_v10_0_resume_mc_access,
};
static void dce_v10_0_set_display_funcs(struct amdgpu_device *adev)

84
drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
View file

@ -499,79 +499,6 @@ static bool dce_v11_0_is_display_hung(struct amdgpu_device *adev)
return true;
}
static void dce_v11_0_stop_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 crtc_enabled, tmp;
int i;
save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
/* disable VGA render */
tmp = RREG32(mmVGA_RENDER_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
WREG32(mmVGA_RENDER_CONTROL, tmp);
/* blank the display controllers */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
CRTC_CONTROL, CRTC_MASTER_EN);
if (crtc_enabled) {
#if 1
save->crtc_enabled[i] = true;
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
/*it is correct only for RGB ; black is 0*/
WREG32(mmCRTC_BLANK_DATA_COLOR + crtc_offsets[i], 0);
tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
}
#else
/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
save->crtc_enabled[i] = false;
/* ***** */
#endif
} else {
save->crtc_enabled[i] = false;
}
}
}
static void dce_v11_0_resume_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 tmp;
int i;
/* update crtc base addresses */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
upper_32_bits(adev->mc.vram_start));
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
(u32)adev->mc.vram_start);
if (save->crtc_enabled[i]) {
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
}
}
WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
WREG32(mmVGA_MEMORY_BASE_ADDRESS, lower_32_bits(adev->mc.vram_start));
/* Unlock vga access */
WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
mdelay(1);
WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
}
static void dce_v11_0_set_vga_render_state(struct amdgpu_device *adev,
bool render)
{
@ -2579,7 +2506,7 @@ static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
aobj = gem_to_amdgpu_bo(obj);
ret = amdgpu_bo_reserve(aobj, false);
if (ret != 0) {
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2587,7 +2514,7 @@ static int dce_v11_0_crtc_cursor_set2(struct drm_crtc *crtc,
amdgpu_bo_unreserve(aobj);
if (ret) {
DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2621,7 +2548,7 @@ unpin:
amdgpu_bo_unpin(aobj);
amdgpu_bo_unreserve(aobj);
}
drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
}
amdgpu_crtc->cursor_bo = obj;
@ -3073,6 +3000,8 @@ static int dce_v11_0_hw_init(void *handle)
dce_v11_0_init_golden_registers(adev);
/* disable vga render */
dce_v11_0_set_vga_render_state(adev, false);
/* init dig PHYs, disp eng pll */
amdgpu_atombios_crtc_powergate_init(adev);
amdgpu_atombios_encoder_init_dig(adev);
@ -3793,7 +3722,6 @@ static void dce_v11_0_encoder_add(struct amdgpu_device *adev,
}
static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
.set_vga_render_state = &dce_v11_0_set_vga_render_state,
.bandwidth_update = &dce_v11_0_bandwidth_update,
.vblank_get_counter = &dce_v11_0_vblank_get_counter,
.vblank_wait = &dce_v11_0_vblank_wait,
@ -3806,8 +3734,6 @@ static const struct amdgpu_display_funcs dce_v11_0_display_funcs = {
.page_flip_get_scanoutpos = &dce_v11_0_crtc_get_scanoutpos,
.add_encoder = &dce_v11_0_encoder_add,
.add_connector = &amdgpu_connector_add,
.stop_mc_access = &dce_v11_0_stop_mc_access,
.resume_mc_access = &dce_v11_0_resume_mc_access,
};
static void dce_v11_0_set_display_funcs(struct amdgpu_device *adev)

123
drivers/gpu/drm/amd/amdgpu/dce_v6_0.c
View file

@ -42,6 +42,7 @@
#include "dce/dce_6_0_d.h"
#include "dce/dce_6_0_sh_mask.h"
#include "gca/gfx_7_2_enum.h"
#include "dce_v6_0.h"
#include "si_enums.h"
static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev);
@ -392,117 +393,6 @@ static u32 dce_v6_0_hpd_get_gpio_reg(struct amdgpu_device *adev)
return mmDC_GPIO_HPD_A;
}
static u32 evergreen_get_vblank_counter(struct amdgpu_device* adev, int crtc)
{
if (crtc >= adev->mode_info.num_crtc)
return 0;
else
return RREG32(mmCRTC_STATUS_FRAME_COUNT + crtc_offsets[crtc]);
}
static void dce_v6_0_stop_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 crtc_enabled, tmp, frame_count;
int i, j;
save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
/* disable VGA render */
WREG32(mmVGA_RENDER_CONTROL, 0);
/* blank the display controllers */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
crtc_enabled = RREG32(mmCRTC_CONTROL + crtc_offsets[i]) & CRTC_CONTROL__CRTC_MASTER_EN_MASK;
if (crtc_enabled) {
save->crtc_enabled[i] = true;
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
if (!(tmp & CRTC_BLANK_CONTROL__CRTC_BLANK_DATA_EN_MASK)) {
dce_v6_0_vblank_wait(adev, i);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
tmp |= CRTC_BLANK_CONTROL__CRTC_BLANK_DATA_EN_MASK;
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
}
/* wait for the next frame */
frame_count = evergreen_get_vblank_counter(adev, i);
for (j = 0; j < adev->usec_timeout; j++) {
if (evergreen_get_vblank_counter(adev, i) != frame_count)
break;
udelay(1);
}
/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
tmp &= ~CRTC_CONTROL__CRTC_MASTER_EN_MASK;
WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
save->crtc_enabled[i] = false;
/* ***** */
} else {
save->crtc_enabled[i] = false;
}
}
}
static void dce_v6_0_resume_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 tmp;
int i, j;
/* update crtc base addresses */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
upper_32_bits(adev->mc.vram_start));
WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
upper_32_bits(adev->mc.vram_start));
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
(u32)adev->mc.vram_start);
WREG32(mmGRPH_SECONDARY_SURFACE_ADDRESS + crtc_offsets[i],
(u32)adev->mc.vram_start);
}
WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
WREG32(mmVGA_MEMORY_BASE_ADDRESS, (u32)adev->mc.vram_start);
/* unlock regs and wait for update */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
if (save->crtc_enabled[i]) {
tmp = RREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i]);
if ((tmp & 0x7) != 0) {
tmp &= ~0x7;
WREG32(mmMASTER_UPDATE_MODE + crtc_offsets[i], tmp);
}
tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
if (tmp & GRPH_UPDATE__GRPH_UPDATE_LOCK_MASK) {
tmp &= ~GRPH_UPDATE__GRPH_UPDATE_LOCK_MASK;
WREG32(mmGRPH_UPDATE + crtc_offsets[i], tmp);
}
tmp = RREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i]);
if (tmp & 1) {
tmp &= ~1;
WREG32(mmMASTER_UPDATE_LOCK + crtc_offsets[i], tmp);
}
for (j = 0; j < adev->usec_timeout; j++) {
tmp = RREG32(mmGRPH_UPDATE + crtc_offsets[i]);
if ((tmp & GRPH_UPDATE__GRPH_SURFACE_UPDATE_PENDING_MASK) == 0)
break;
udelay(1);
}
}
}
/* Unlock vga access */
WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
mdelay(1);
WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
}
static void dce_v6_0_set_vga_render_state(struct amdgpu_device *adev,
bool render)
{
@ -2432,7 +2322,7 @@ static int dce_v6_0_crtc_cursor_set2(struct drm_crtc *crtc,
aobj = gem_to_amdgpu_bo(obj);
ret = amdgpu_bo_reserve(aobj, false);
if (ret != 0) {
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2440,7 +2330,7 @@ static int dce_v6_0_crtc_cursor_set2(struct drm_crtc *crtc,
amdgpu_bo_unreserve(aobj);
if (ret) {
DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2474,7 +2364,7 @@ unpin:
amdgpu_bo_unpin(aobj);
amdgpu_bo_unreserve(aobj);
}
drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
}
amdgpu_crtc->cursor_bo = obj;
@ -2860,6 +2750,8 @@ static int dce_v6_0_hw_init(void *handle)
int i;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
/* disable vga render */
dce_v6_0_set_vga_render_state(adev, false);
/* init dig PHYs, disp eng pll */
amdgpu_atombios_encoder_init_dig(adev);
amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
@ -3512,7 +3404,6 @@ static void dce_v6_0_encoder_add(struct amdgpu_device *adev,
}
static const struct amdgpu_display_funcs dce_v6_0_display_funcs = {
.set_vga_render_state = &dce_v6_0_set_vga_render_state,
.bandwidth_update = &dce_v6_0_bandwidth_update,
.vblank_get_counter = &dce_v6_0_vblank_get_counter,
.vblank_wait = &dce_v6_0_vblank_wait,
@ -3525,8 +3416,6 @@ static const struct amdgpu_display_funcs dce_v6_0_display_funcs = {
.page_flip_get_scanoutpos = &dce_v6_0_crtc_get_scanoutpos,
.add_encoder = &dce_v6_0_encoder_add,
.add_connector = &amdgpu_connector_add,
.stop_mc_access = &dce_v6_0_stop_mc_access,
.resume_mc_access = &dce_v6_0_resume_mc_access,
};
static void dce_v6_0_set_display_funcs(struct amdgpu_device *adev)

86
drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
View file

@ -419,81 +419,6 @@ static bool dce_v8_0_is_display_hung(struct amdgpu_device *adev)
return true;
}
static void dce_v8_0_stop_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 crtc_enabled, tmp;
int i;
save->vga_render_control = RREG32(mmVGA_RENDER_CONTROL);
save->vga_hdp_control = RREG32(mmVGA_HDP_CONTROL);
/* disable VGA render */
tmp = RREG32(mmVGA_RENDER_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
WREG32(mmVGA_RENDER_CONTROL, tmp);
/* blank the display controllers */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
crtc_enabled = REG_GET_FIELD(RREG32(mmCRTC_CONTROL + crtc_offsets[i]),
CRTC_CONTROL, CRTC_MASTER_EN);
if (crtc_enabled) {
#if 1
save->crtc_enabled[i] = true;
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
if (REG_GET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN) == 0) {
/*it is correct only for RGB ; black is 0*/
WREG32(mmCRTC_BLANK_DATA_COLOR + crtc_offsets[i], 0);
tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 1);
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
}
mdelay(20);
#else
/* XXX this is a hack to avoid strange behavior with EFI on certain systems */
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 1);
tmp = RREG32(mmCRTC_CONTROL + crtc_offsets[i]);
tmp = REG_SET_FIELD(tmp, CRTC_CONTROL, CRTC_MASTER_EN, 0);
WREG32(mmCRTC_CONTROL + crtc_offsets[i], tmp);
WREG32(mmCRTC_UPDATE_LOCK + crtc_offsets[i], 0);
save->crtc_enabled[i] = false;
/* ***** */
#endif
} else {
save->crtc_enabled[i] = false;
}
}
}
static void dce_v8_0_resume_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
u32 tmp;
int i;
/* update crtc base addresses */
for (i = 0; i < adev->mode_info.num_crtc; i++) {
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS_HIGH + crtc_offsets[i],
upper_32_bits(adev->mc.vram_start));
WREG32(mmGRPH_PRIMARY_SURFACE_ADDRESS + crtc_offsets[i],
(u32)adev->mc.vram_start);
if (save->crtc_enabled[i]) {
tmp = RREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i]);
tmp = REG_SET_FIELD(tmp, CRTC_BLANK_CONTROL, CRTC_BLANK_DATA_EN, 0);
WREG32(mmCRTC_BLANK_CONTROL + crtc_offsets[i], tmp);
}
mdelay(20);
}
WREG32(mmVGA_MEMORY_BASE_ADDRESS_HIGH, upper_32_bits(adev->mc.vram_start));
WREG32(mmVGA_MEMORY_BASE_ADDRESS, lower_32_bits(adev->mc.vram_start));
/* Unlock vga access */
WREG32(mmVGA_HDP_CONTROL, save->vga_hdp_control);
mdelay(1);
WREG32(mmVGA_RENDER_CONTROL, save->vga_render_control);
}
static void dce_v8_0_set_vga_render_state(struct amdgpu_device *adev,
bool render)
{
@ -2410,7 +2335,7 @@ static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
aobj = gem_to_amdgpu_bo(obj);
ret = amdgpu_bo_reserve(aobj, false);
if (ret != 0) {
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2418,7 +2343,7 @@ static int dce_v8_0_crtc_cursor_set2(struct drm_crtc *crtc,
amdgpu_bo_unreserve(aobj);
if (ret) {
DRM_ERROR("Failed to pin new cursor BO (%d)\n", ret);
drm_gem_object_unreference_unlocked(obj);
drm_gem_object_put_unlocked(obj);
return ret;
}
@ -2452,7 +2377,7 @@ unpin:
amdgpu_bo_unpin(aobj);
amdgpu_bo_unreserve(aobj);
}
drm_gem_object_unreference_unlocked(amdgpu_crtc->cursor_bo);
drm_gem_object_put_unlocked(amdgpu_crtc->cursor_bo);
}
amdgpu_crtc->cursor_bo = obj;
@ -2857,6 +2782,8 @@ static int dce_v8_0_hw_init(void *handle)
int i;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
/* disable vga render */
dce_v8_0_set_vga_render_state(adev, false);
/* init dig PHYs, disp eng pll */
amdgpu_atombios_encoder_init_dig(adev);
amdgpu_atombios_crtc_set_disp_eng_pll(adev, adev->clock.default_dispclk);
@ -3561,7 +3488,6 @@ static void dce_v8_0_encoder_add(struct amdgpu_device *adev,
}
static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
.set_vga_render_state = &dce_v8_0_set_vga_render_state,
.bandwidth_update = &dce_v8_0_bandwidth_update,
.vblank_get_counter = &dce_v8_0_vblank_get_counter,
.vblank_wait = &dce_v8_0_vblank_wait,
@ -3574,8 +3500,6 @@ static const struct amdgpu_display_funcs dce_v8_0_display_funcs = {
.page_flip_get_scanoutpos = &dce_v8_0_crtc_get_scanoutpos,
.add_encoder = &dce_v8_0_encoder_add,
.add_connector = &amdgpu_connector_add,
.stop_mc_access = &dce_v8_0_stop_mc_access,
.resume_mc_access = &dce_v8_0_resume_mc_access,
};
static void dce_v8_0_set_display_funcs(struct amdgpu_device *adev)

102
drivers/gpu/drm/amd/amdgpu/dce_virtual.c
View file

@ -95,62 +95,6 @@ static u32 dce_virtual_hpd_get_gpio_reg(struct amdgpu_device *adev)
return 0;
}
static void dce_virtual_stop_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
switch (adev->asic_type) {
#ifdef CONFIG_DRM_AMDGPU_SI
case CHIP_TAHITI:
case CHIP_PITCAIRN:
case CHIP_VERDE:
case CHIP_OLAND:
dce_v6_0_disable_dce(adev);
break;
#endif
#ifdef CONFIG_DRM_AMDGPU_CIK
case CHIP_BONAIRE:
case CHIP_HAWAII:
case CHIP_KAVERI:
case CHIP_KABINI:
case CHIP_MULLINS:
dce_v8_0_disable_dce(adev);
break;
#endif
case CHIP_FIJI:
case CHIP_TONGA:
dce_v10_0_disable_dce(adev);
break;
case CHIP_CARRIZO:
case CHIP_STONEY:
case CHIP_POLARIS10:
case CHIP_POLARIS11:
case CHIP_POLARIS12:
dce_v11_0_disable_dce(adev);
break;
case CHIP_TOPAZ:
#ifdef CONFIG_DRM_AMDGPU_SI
case CHIP_HAINAN:
#endif
/* no DCE */
return;
default:
DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type);
}
return;
}
static void dce_virtual_resume_mc_access(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
{
return;
}
static void dce_virtual_set_vga_render_state(struct amdgpu_device *adev,
bool render)
{
return;
}
/**
* dce_virtual_bandwidth_update - program display watermarks
*
@ -499,6 +443,47 @@ static int dce_virtual_sw_fini(void *handle)
static int dce_virtual_hw_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
switch (adev->asic_type) {
#ifdef CONFIG_DRM_AMDGPU_SI
case CHIP_TAHITI:
case CHIP_PITCAIRN:
case CHIP_VERDE:
case CHIP_OLAND:
dce_v6_0_disable_dce(adev);
break;
#endif
#ifdef CONFIG_DRM_AMDGPU_CIK
case CHIP_BONAIRE:
case CHIP_HAWAII:
case CHIP_KAVERI:
case CHIP_KABINI:
case CHIP_MULLINS:
dce_v8_0_disable_dce(adev);
break;
#endif
case CHIP_FIJI:
case CHIP_TONGA:
dce_v10_0_disable_dce(adev);
break;
case CHIP_CARRIZO:
case CHIP_STONEY:
case CHIP_POLARIS11:
case CHIP_POLARIS10:
dce_v11_0_disable_dce(adev);
break;
case CHIP_TOPAZ:
#ifdef CONFIG_DRM_AMDGPU_SI
case CHIP_HAINAN:
#endif
/* no DCE */
break;
case CHIP_VEGA10:
break;
default:
DRM_ERROR("Virtual display unsupported ASIC type: 0x%X\n", adev->asic_type);
}
return 0;
}
@ -654,7 +639,6 @@ static int dce_virtual_connector_encoder_init(struct amdgpu_device *adev,
}
static const struct amdgpu_display_funcs dce_virtual_display_funcs = {
.set_vga_render_state = &dce_virtual_set_vga_render_state,
.bandwidth_update = &dce_virtual_bandwidth_update,
.vblank_get_counter = &dce_virtual_vblank_get_counter,
.vblank_wait = &dce_virtual_vblank_wait,
@ -667,8 +651,6 @@ static const struct amdgpu_display_funcs dce_virtual_display_funcs = {
.page_flip_get_scanoutpos = &dce_virtual_crtc_get_scanoutpos,
.add_encoder = NULL,
.add_connector = NULL,
.stop_mc_access = &dce_virtual_stop_mc_access,
.resume_mc_access = &dce_virtual_resume_mc_access,
};
static void dce_virtual_set_display_funcs(struct amdgpu_device *adev)
@ -786,7 +768,7 @@ static const struct amdgpu_irq_src_funcs dce_virtual_crtc_irq_funcs = {
static void dce_virtual_set_irq_funcs(struct amdgpu_device *adev)
{
adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_LAST;
adev->crtc_irq.num_types = AMDGPU_CRTC_IRQ_VBLANK6 + 1;
adev->crtc_irq.funcs = &dce_virtual_crtc_irq_funcs;
}

109
drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
View file

@ -1573,7 +1573,7 @@ static void gfx_v6_0_gpu_init(struct amdgpu_device *adev)
static void gfx_v6_0_scratch_init(struct amdgpu_device *adev)
{
adev->gfx.scratch.num_reg = 7;
adev->gfx.scratch.num_reg = 8;
adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
}
@ -2217,40 +2217,9 @@ static void gfx_v6_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
static void gfx_v6_0_rlc_fini(struct amdgpu_device *adev)
{
int r;
if (adev->gfx.rlc.save_restore_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC sr bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.save_restore_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
amdgpu_bo_unref(&adev->gfx.rlc.save_restore_obj);
adev->gfx.rlc.save_restore_obj = NULL;
}
if (adev->gfx.rlc.clear_state_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC c bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
adev->gfx.rlc.clear_state_obj = NULL;
}
if (adev->gfx.rlc.cp_table_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.cp_table_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
amdgpu_bo_unref(&adev->gfx.rlc.cp_table_obj);
adev->gfx.rlc.cp_table_obj = NULL;
}
amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
}
static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
@ -2273,43 +2242,23 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
if (src_ptr) {
/* save restore block */
if (adev->gfx.rlc.save_restore_obj == NULL) {
r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
NULL, NULL,
&adev->gfx.rlc.save_restore_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, false);
if (unlikely(r != 0)) {
gfx_v6_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.save_restore_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.save_restore_gpu_addr);
r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.save_restore_obj,
&adev->gfx.rlc.save_restore_gpu_addr,
(void **)&adev->gfx.rlc.sr_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
dev_warn(adev->dev, "(%d) pin RLC sr bo failed\n", r);
dev_warn(adev->dev, "(%d) create RLC sr bo failed\n",
r);
gfx_v6_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.save_restore_obj, (void **)&adev->gfx.rlc.sr_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC sr bo failed\n", r);
gfx_v6_0_rlc_fini(adev);
return r;
}
/* write the sr buffer */
dst_ptr = adev->gfx.rlc.sr_ptr;
for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
dst_ptr[i] = cpu_to_le32(src_ptr[i]);
amdgpu_bo_kunmap(adev->gfx.rlc.save_restore_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
}
@ -2319,39 +2268,17 @@ static int gfx_v6_0_rlc_init(struct amdgpu_device *adev)
adev->gfx.rlc.clear_state_size = gfx_v6_0_get_csb_size(adev);
dws = adev->gfx.rlc.clear_state_size + (256 / 4);
if (adev->gfx.rlc.clear_state_obj == NULL) {
r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED,
NULL, NULL,
&adev->gfx.rlc.clear_state_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
gfx_v6_0_rlc_fini(adev);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
if (unlikely(r != 0)) {
gfx_v6_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_gpu_addr);
r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_obj,
&adev->gfx.rlc.clear_state_gpu_addr,
(void **)&adev->gfx.rlc.cs_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
dev_warn(adev->dev, "(%d) pin RLC c bo failed\n", r);
dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
gfx_v6_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC c bo failed\n", r);
gfx_v6_0_rlc_fini(adev);
return r;
}
/* set up the cs buffer */
dst_ptr = adev->gfx.rlc.cs_ptr;
reg_list_mc_addr = adev->gfx.rlc.clear_state_gpu_addr + 256;

241
drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
View file

@ -1823,7 +1823,7 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
}
/**
* gmc_v7_0_init_compute_vmid - gart enable
* gfx_v7_0_init_compute_vmid - gart enable
*
* @adev: amdgpu_device pointer
*
@ -1833,7 +1833,7 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
#define DEFAULT_SH_MEM_BASES (0x6000)
#define FIRST_COMPUTE_VMID (8)
#define LAST_COMPUTE_VMID (16)
static void gmc_v7_0_init_compute_vmid(struct amdgpu_device *adev)
static void gfx_v7_0_init_compute_vmid(struct amdgpu_device *adev)
{
int i;
uint32_t sh_mem_config;
@ -1939,7 +1939,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
cik_srbm_select(adev, 0, 0, 0, 0);
mutex_unlock(&adev->srbm_mutex);
gmc_v7_0_init_compute_vmid(adev);
gfx_v7_0_init_compute_vmid(adev);
WREG32(mmSX_DEBUG_1, 0x20);
@ -2021,7 +2021,7 @@ static void gfx_v7_0_gpu_init(struct amdgpu_device *adev)
*/
static void gfx_v7_0_scratch_init(struct amdgpu_device *adev)
{
adev->gfx.scratch.num_reg = 7;
adev->gfx.scratch.num_reg = 8;
adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
}
@ -2774,39 +2774,18 @@ static int gfx_v7_0_cp_compute_load_microcode(struct amdgpu_device *adev)
*/
static void gfx_v7_0_cp_compute_fini(struct amdgpu_device *adev)
{
int i, r;
int i;
for (i = 0; i < adev->gfx.num_compute_rings; i++) {
struct amdgpu_ring *ring = &adev->gfx.compute_ring[i];
if (ring->mqd_obj) {
r = amdgpu_bo_reserve(ring->mqd_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve MQD bo failed\n", r);
amdgpu_bo_unpin(ring->mqd_obj);
amdgpu_bo_unreserve(ring->mqd_obj);
amdgpu_bo_unref(&ring->mqd_obj);
ring->mqd_obj = NULL;
}
amdgpu_bo_free_kernel(&ring->mqd_obj, NULL, NULL);
}
}
static void gfx_v7_0_mec_fini(struct amdgpu_device *adev)
{
int r;
if (adev->gfx.mec.hpd_eop_obj) {
r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
adev->gfx.mec.hpd_eop_obj = NULL;
}
amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
}
static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
@ -2823,33 +2802,14 @@ static int gfx_v7_0_mec_init(struct amdgpu_device *adev)
/* allocate space for ALL pipes (even the ones we don't own) */
mec_hpd_size = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec
* GFX7_MEC_HPD_SIZE * 2;
if (adev->gfx.mec.hpd_eop_obj == NULL) {
r = amdgpu_bo_create(adev,
mec_hpd_size,
PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
&adev->gfx.mec.hpd_eop_obj);
if (r) {
dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
if (unlikely(r != 0)) {
gfx_v7_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.hpd_eop_gpu_addr);
r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.hpd_eop_obj,
&adev->gfx.mec.hpd_eop_gpu_addr,
(void **)&hpd);
if (r) {
dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
gfx_v7_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
if (r) {
dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
dev_warn(adev->dev, "(%d) create, pin or map of HDP EOP bo failed\n", r);
gfx_v7_0_mec_fini(adev);
return r;
}
@ -3108,32 +3068,12 @@ static int gfx_v7_0_compute_queue_init(struct amdgpu_device *adev, int ring_id)
struct cik_mqd *mqd;
struct amdgpu_ring *ring = &adev->gfx.compute_ring[ring_id];
if (ring->mqd_obj == NULL) {
r = amdgpu_bo_create(adev,
sizeof(struct cik_mqd),
PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
&ring->mqd_obj);
if (r) {
dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(ring->mqd_obj, false);
if (unlikely(r != 0))
goto out;
r = amdgpu_bo_pin(ring->mqd_obj, AMDGPU_GEM_DOMAIN_GTT,
&mqd_gpu_addr);
r = amdgpu_bo_create_reserved(adev, sizeof(struct cik_mqd), PAGE_SIZE,
AMDGPU_GEM_DOMAIN_GTT, &ring->mqd_obj,
&mqd_gpu_addr, (void **)&mqd);
if (r) {
dev_warn(adev->dev, "(%d) pin MQD bo failed\n", r);
goto out_unreserve;
}
r = amdgpu_bo_kmap(ring->mqd_obj, (void **)&mqd);
if (r) {
dev_warn(adev->dev, "(%d) map MQD bo failed\n", r);
goto out_unreserve;
dev_warn(adev->dev, "(%d) create MQD bo failed\n", r);
return r;
}
mutex_lock(&adev->srbm_mutex);
@ -3147,9 +3087,7 @@ static int gfx_v7_0_compute_queue_init(struct amdgpu_device *adev, int ring_id)
mutex_unlock(&adev->srbm_mutex);
amdgpu_bo_kunmap(ring->mqd_obj);
out_unreserve:
amdgpu_bo_unreserve(ring->mqd_obj);
out:
return 0;
}
@ -3361,43 +3299,9 @@ static void gfx_v7_0_ring_emit_vm_flush(struct amdgpu_ring *ring,
*/
static void gfx_v7_0_rlc_fini(struct amdgpu_device *adev)
{
int r;
/* save restore block */
if (adev->gfx.rlc.save_restore_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC sr bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.save_restore_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
amdgpu_bo_unref(&adev->gfx.rlc.save_restore_obj);
adev->gfx.rlc.save_restore_obj = NULL;
}
/* clear state block */
if (adev->gfx.rlc.clear_state_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC c bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
adev->gfx.rlc.clear_state_obj = NULL;
}
/* clear state block */
if (adev->gfx.rlc.cp_table_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.cp_table_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
amdgpu_bo_unref(&adev->gfx.rlc.cp_table_obj);
adev->gfx.rlc.cp_table_obj = NULL;
}
amdgpu_bo_free_kernel(&adev->gfx.rlc.save_restore_obj, NULL, NULL);
amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
}
static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
@ -3432,39 +3336,17 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
if (src_ptr) {
/* save restore block */
if (adev->gfx.rlc.save_restore_obj == NULL) {
r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL,
&adev->gfx.rlc.save_restore_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC sr bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.save_restore_obj, false);
if (unlikely(r != 0)) {
gfx_v7_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.save_restore_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.save_restore_gpu_addr);
r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.save_restore_obj,
&adev->gfx.rlc.save_restore_gpu_addr,
(void **)&adev->gfx.rlc.sr_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.save_restore_obj);
dev_warn(adev->dev, "(%d) pin RLC sr bo failed\n", r);
dev_warn(adev->dev, "(%d) create, pin or map of RLC sr bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.save_restore_obj, (void **)&adev->gfx.rlc.sr_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC sr bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
/* write the sr buffer */
dst_ptr = adev->gfx.rlc.sr_ptr;
for (i = 0; i < adev->gfx.rlc.reg_list_size; i++)
@ -3477,39 +3359,17 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
/* clear state block */
adev->gfx.rlc.clear_state_size = dws = gfx_v7_0_get_csb_size(adev);
if (adev->gfx.rlc.clear_state_obj == NULL) {
r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL,
&adev->gfx.rlc.clear_state_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
if (unlikely(r != 0)) {
gfx_v7_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_gpu_addr);
r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_obj,
&adev->gfx.rlc.clear_state_gpu_addr,
(void **)&adev->gfx.rlc.cs_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
dev_warn(adev->dev, "(%d) pin RLC c bo failed\n", r);
dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC c bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
/* set up the cs buffer */
dst_ptr = adev->gfx.rlc.cs_ptr;
gfx_v7_0_get_csb_buffer(adev, dst_ptr);
@ -3518,37 +3378,14 @@ static int gfx_v7_0_rlc_init(struct amdgpu_device *adev)
}
if (adev->gfx.rlc.cp_table_size) {
if (adev->gfx.rlc.cp_table_obj == NULL) {
r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL,
&adev->gfx.rlc.cp_table_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, false);
if (unlikely(r != 0)) {
dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.cp_table_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.cp_table_gpu_addr);
r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.cp_table_obj,
&adev->gfx.rlc.cp_table_gpu_addr,
(void **)&adev->gfx.rlc.cp_table_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
dev_warn(adev->dev, "(%d) pin RLC cp_table bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.cp_table_obj, (void **)&adev->gfx.rlc.cp_table_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC cp table bo failed\n", r);
dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
gfx_v7_0_rlc_fini(adev);
return r;
}

169
drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
View file

@ -193,8 +193,8 @@ static const u32 tonga_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF
};
static const u32 tonga_mgcg_cgcg_init[] =
@ -303,8 +303,8 @@ static const u32 polaris11_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x22011002,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
};
static const u32 golden_settings_polaris10_a11[] =
@ -336,8 +336,8 @@ static const u32 polaris10_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
};
static const u32 fiji_golden_common_all[] =
@ -348,8 +348,8 @@ static const u32 fiji_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x22011003,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
mmGRBM_GFX_INDEX, 0xffffffff, 0xe0000000,
mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x00000009,
};
@ -436,8 +436,8 @@ static const u32 iceland_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x22010001,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF
};
static const u32 iceland_mgcg_cgcg_init[] =
@ -532,8 +532,8 @@ static const u32 cz_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x22010001,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF
};
static const u32 cz_mgcg_cgcg_init[] =
@ -637,8 +637,8 @@ static const u32 stoney_golden_common_all[] =
mmGB_ADDR_CONFIG, 0xffffffff, 0x12010001,
mmSPI_RESOURCE_RESERVE_CU_0, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_CU_1, 0xffffffff, 0x00000800,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00007FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00007FAF,
mmSPI_RESOURCE_RESERVE_EN_CU_0, 0xffffffff, 0x00FF7FBF,
mmSPI_RESOURCE_RESERVE_EN_CU_1, 0xffffffff, 0x00FF7FAF,
};
static const u32 stoney_mgcg_cgcg_init[] =
@ -750,7 +750,7 @@ static void gfx_v8_0_init_golden_registers(struct amdgpu_device *adev)
static void gfx_v8_0_scratch_init(struct amdgpu_device *adev)
{
adev->gfx.scratch.num_reg = 7;
adev->gfx.scratch.num_reg = 8;
adev->gfx.scratch.reg_base = mmSCRATCH_REG0;
adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
}
@ -1238,29 +1238,8 @@ static void cz_init_cp_jump_table(struct amdgpu_device *adev)
static void gfx_v8_0_rlc_fini(struct amdgpu_device *adev)
{
int r;
/* clear state block */
if (adev->gfx.rlc.clear_state_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC cbs bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.clear_state_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
amdgpu_bo_unref(&adev->gfx.rlc.clear_state_obj);
adev->gfx.rlc.clear_state_obj = NULL;
}
/* jump table block */
if (adev->gfx.rlc.cp_table_obj) {
r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.rlc.cp_table_obj);
amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
amdgpu_bo_unref(&adev->gfx.rlc.cp_table_obj);
adev->gfx.rlc.cp_table_obj = NULL;
}
amdgpu_bo_free_kernel(&adev->gfx.rlc.clear_state_obj, NULL, NULL);
amdgpu_bo_free_kernel(&adev->gfx.rlc.cp_table_obj, NULL, NULL);
}
static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
@ -1278,39 +1257,17 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
/* clear state block */
adev->gfx.rlc.clear_state_size = dws = gfx_v8_0_get_csb_size(adev);
if (adev->gfx.rlc.clear_state_obj == NULL) {
r = amdgpu_bo_create(adev, dws * 4, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL,
&adev->gfx.rlc.clear_state_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
gfx_v8_0_rlc_fini(adev);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.clear_state_obj, false);
if (unlikely(r != 0)) {
gfx_v8_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.clear_state_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_gpu_addr);
r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_obj,
&adev->gfx.rlc.clear_state_gpu_addr,
(void **)&adev->gfx.rlc.cs_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.clear_state_obj);
dev_warn(adev->dev, "(%d) pin RLC cbs bo failed\n", r);
dev_warn(adev->dev, "(%d) create RLC c bo failed\n", r);
gfx_v8_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.clear_state_obj, (void **)&adev->gfx.rlc.cs_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC cbs bo failed\n", r);
gfx_v8_0_rlc_fini(adev);
return r;
}
/* set up the cs buffer */
dst_ptr = adev->gfx.rlc.cs_ptr;
gfx_v8_0_get_csb_buffer(adev, dst_ptr);
@ -1321,34 +1278,13 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
if ((adev->asic_type == CHIP_CARRIZO) ||
(adev->asic_type == CHIP_STONEY)) {
adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
if (adev->gfx.rlc.cp_table_obj == NULL) {
r = amdgpu_bo_create(adev, adev->gfx.rlc.cp_table_size, PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_VRAM,
AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED |
AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS,
NULL, NULL,
&adev->gfx.rlc.cp_table_obj);
if (r) {
dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.rlc.cp_table_obj, false);
if (unlikely(r != 0)) {
dev_warn(adev->dev, "(%d) reserve RLC cp table bo failed\n", r);
return r;
}
r = amdgpu_bo_pin(adev->gfx.rlc.cp_table_obj, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.cp_table_gpu_addr);
r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.cp_table_obj,
&adev->gfx.rlc.cp_table_gpu_addr,
(void **)&adev->gfx.rlc.cp_table_ptr);
if (r) {
amdgpu_bo_unreserve(adev->gfx.rlc.cp_table_obj);
dev_warn(adev->dev, "(%d) pin RLC cp table bo failed\n", r);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.rlc.cp_table_obj, (void **)&adev->gfx.rlc.cp_table_ptr);
if (r) {
dev_warn(adev->dev, "(%d) map RLC cp table bo failed\n", r);
dev_warn(adev->dev, "(%d) create RLC cp table bo failed\n", r);
return r;
}
@ -1363,17 +1299,7 @@ static int gfx_v8_0_rlc_init(struct amdgpu_device *adev)
static void gfx_v8_0_mec_fini(struct amdgpu_device *adev)
{
int r;
if (adev->gfx.mec.hpd_eop_obj) {
r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
adev->gfx.mec.hpd_eop_obj = NULL;
}
amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
}
static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
@ -1389,34 +1315,13 @@ static int gfx_v8_0_mec_init(struct amdgpu_device *adev)
mec_hpd_size = adev->gfx.num_compute_rings * GFX8_MEC_HPD_SIZE;
if (adev->gfx.mec.hpd_eop_obj == NULL) {
r = amdgpu_bo_create(adev,
mec_hpd_size,
PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
&adev->gfx.mec.hpd_eop_obj);
if (r) {
dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
if (unlikely(r != 0)) {
gfx_v8_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.hpd_eop_gpu_addr);
r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.hpd_eop_obj,
&adev->gfx.mec.hpd_eop_gpu_addr,
(void **)&hpd);
if (r) {
dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
gfx_v8_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
if (r) {
dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
gfx_v8_0_mec_fini(adev);
dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
return r;
}
@ -4564,7 +4469,7 @@ static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
/* 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))) {
if (WARN_ON(i >= (sizeof(queue_mask)*8))) {
DRM_ERROR("Invalid KCQ enabled: %d\n", i);
break;
}

170
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
View file

@ -116,7 +116,9 @@ static const u32 golden_settings_gc_9_0[] =
SOC15_REG_OFFSET(GC, 0, mmRLC_GPM_UTCL1_CNTL_2), 0x08000000, 0x08000080,
SOC15_REG_OFFSET(GC, 0, mmRLC_PREWALKER_UTCL1_CNTL), 0x08000000, 0x08000080,
SOC15_REG_OFFSET(GC, 0, mmRLC_SPM_UTCL1_CNTL), 0x08000000, 0x08000080,
SOC15_REG_OFFSET(GC, 0, mmSH_MEM_CONFIG), 0x00001000, 0x00001000,
SOC15_REG_OFFSET(GC, 0, mmSPI_CONFIG_CNTL_1), 0x0000000f, 0x01000107,
SOC15_REG_OFFSET(GC, 0, mmSQC_CONFIG), 0x03000000, 0x020a2000,
SOC15_REG_OFFSET(GC, 0, mmTA_CNTL_AUX), 0xfffffeef, 0x010b0000,
SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_HI), 0xffffffff, 0x4a2c0e68,
SOC15_REG_OFFSET(GC, 0, mmTCP_CHAN_STEER_LO), 0xffffffff, 0xb5d3f197,
@ -211,7 +213,7 @@ static void gfx_v9_0_init_golden_registers(struct amdgpu_device *adev)
static void gfx_v9_0_scratch_init(struct amdgpu_device *adev)
{
adev->gfx.scratch.num_reg = 7;
adev->gfx.scratch.num_reg = 8;
adev->gfx.scratch.reg_base = SOC15_REG_OFFSET(GC, 0, mmSCRATCH_REG0);
adev->gfx.scratch.free_mask = (1u << adev->gfx.scratch.num_reg) - 1;
}
@ -772,18 +774,16 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
if (cs_data) {
/* clear state block */
adev->gfx.rlc.clear_state_size = dws = gfx_v9_0_get_csb_size(adev);
if (adev->gfx.rlc.clear_state_obj == NULL) {
r = amdgpu_bo_create_kernel(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_obj,
&adev->gfx.rlc.clear_state_gpu_addr,
(void **)&adev->gfx.rlc.cs_ptr);
if (r) {
dev_err(adev->dev,
"(%d) failed to create rlc csb bo\n", r);
gfx_v9_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_create_reserved(adev, dws * 4, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.clear_state_obj,
&adev->gfx.rlc.clear_state_gpu_addr,
(void **)&adev->gfx.rlc.cs_ptr);
if (r) {
dev_err(adev->dev, "(%d) failed to create rlc csb bo\n",
r);
gfx_v9_0_rlc_fini(adev);
return r;
}
/* set up the cs buffer */
dst_ptr = adev->gfx.rlc.cs_ptr;
@ -795,18 +795,16 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
if (adev->asic_type == CHIP_RAVEN) {
/* TODO: double check the cp_table_size for RV */
adev->gfx.rlc.cp_table_size = ALIGN(96 * 5 * 4, 2048) + (64 * 1024); /* JT + GDS */
if (adev->gfx.rlc.cp_table_obj == NULL) {
r = amdgpu_bo_create_kernel(adev, adev->gfx.rlc.cp_table_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.cp_table_obj,
&adev->gfx.rlc.cp_table_gpu_addr,
(void **)&adev->gfx.rlc.cp_table_ptr);
if (r) {
dev_err(adev->dev,
"(%d) failed to create cp table bo\n", r);
gfx_v9_0_rlc_fini(adev);
return r;
}
r = amdgpu_bo_create_reserved(adev, adev->gfx.rlc.cp_table_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
&adev->gfx.rlc.cp_table_obj,
&adev->gfx.rlc.cp_table_gpu_addr,
(void **)&adev->gfx.rlc.cp_table_ptr);
if (r) {
dev_err(adev->dev,
"(%d) failed to create cp table bo\n", r);
gfx_v9_0_rlc_fini(adev);
return r;
}
rv_init_cp_jump_table(adev);
@ -821,28 +819,8 @@ static int gfx_v9_0_rlc_init(struct amdgpu_device *adev)
static void gfx_v9_0_mec_fini(struct amdgpu_device *adev)
{
int r;
if (adev->gfx.mec.hpd_eop_obj) {
r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve HPD EOP bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.mec.hpd_eop_obj);
amdgpu_bo_unreserve(adev->gfx.mec.hpd_eop_obj);
amdgpu_bo_unref(&adev->gfx.mec.hpd_eop_obj);
adev->gfx.mec.hpd_eop_obj = NULL;
}
if (adev->gfx.mec.mec_fw_obj) {
r = amdgpu_bo_reserve(adev->gfx.mec.mec_fw_obj, true);
if (unlikely(r != 0))
dev_warn(adev->dev, "(%d) reserve mec firmware bo failed\n", r);
amdgpu_bo_unpin(adev->gfx.mec.mec_fw_obj);
amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
amdgpu_bo_unref(&adev->gfx.mec.mec_fw_obj);
adev->gfx.mec.mec_fw_obj = NULL;
}
amdgpu_bo_free_kernel(&adev->gfx.mec.hpd_eop_obj, NULL, NULL);
amdgpu_bo_free_kernel(&adev->gfx.mec.mec_fw_obj, NULL, NULL);
}
static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
@ -862,33 +840,13 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
amdgpu_gfx_compute_queue_acquire(adev);
mec_hpd_size = adev->gfx.num_compute_rings * GFX9_MEC_HPD_SIZE;
if (adev->gfx.mec.hpd_eop_obj == NULL) {
r = amdgpu_bo_create(adev,
mec_hpd_size,
PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
&adev->gfx.mec.hpd_eop_obj);
if (r) {
dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
return r;
}
}
r = amdgpu_bo_reserve(adev->gfx.mec.hpd_eop_obj, false);
if (unlikely(r != 0)) {
gfx_v9_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.mec.hpd_eop_obj, AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.hpd_eop_gpu_addr);
r = amdgpu_bo_create_reserved(adev, mec_hpd_size, PAGE_SIZE,
AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.hpd_eop_obj,
&adev->gfx.mec.hpd_eop_gpu_addr,
(void **)&hpd);
if (r) {
dev_warn(adev->dev, "(%d) pin HDP EOP bo failed\n", r);
gfx_v9_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.mec.hpd_eop_obj, (void **)&hpd);
if (r) {
dev_warn(adev->dev, "(%d) map HDP EOP bo failed\n", r);
dev_warn(adev->dev, "(%d) create HDP EOP bo failed\n", r);
gfx_v9_0_mec_fini(adev);
return r;
}
@ -905,42 +863,22 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
le32_to_cpu(mec_hdr->header.ucode_array_offset_bytes));
fw_size = le32_to_cpu(mec_hdr->header.ucode_size_bytes) / 4;
if (adev->gfx.mec.mec_fw_obj == NULL) {
r = amdgpu_bo_create(adev,
mec_hdr->header.ucode_size_bytes,
PAGE_SIZE, true,
AMDGPU_GEM_DOMAIN_GTT, 0, NULL, NULL,
&adev->gfx.mec.mec_fw_obj);
if (r) {
dev_warn(adev->dev, "(%d) create mec firmware bo failed\n", r);
return r;
}
r = amdgpu_bo_create_reserved(adev, mec_hdr->header.ucode_size_bytes,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.mec_fw_obj,
&adev->gfx.mec.mec_fw_gpu_addr,
(void **)&fw);
if (r) {
dev_warn(adev->dev, "(%d) create mec firmware bo failed\n", r);
gfx_v9_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_reserve(adev->gfx.mec.mec_fw_obj, false);
if (unlikely(r != 0)) {
gfx_v9_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_pin(adev->gfx.mec.mec_fw_obj, AMDGPU_GEM_DOMAIN_GTT,
&adev->gfx.mec.mec_fw_gpu_addr);
if (r) {
dev_warn(adev->dev, "(%d) pin mec firmware bo failed\n", r);
gfx_v9_0_mec_fini(adev);
return r;
}
r = amdgpu_bo_kmap(adev->gfx.mec.mec_fw_obj, (void **)&fw);
if (r) {
dev_warn(adev->dev, "(%d) map firmware bo failed\n", r);
gfx_v9_0_mec_fini(adev);
return r;
}
memcpy(fw, fw_data, fw_size);
amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj);
amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj);
return 0;
}
@ -1475,21 +1413,23 @@ static void gfx_v9_0_tiling_mode_table_init(struct amdgpu_device *adev)
static void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num, u32 instance)
{
u32 data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
u32 data;
if ((se_num == 0xffffffff) && (sh_num == 0xffffffff)) {
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
if (instance == 0xffffffff)
data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES, 1);
else
data = REG_SET_FIELD(0, GRBM_GFX_INDEX, INSTANCE_INDEX, instance);
if (se_num == 0xffffffff)
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1);
} else if (se_num == 0xffffffff) {
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_BROADCAST_WRITES, 1);
} else if (sh_num == 0xffffffff) {
else
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
if (sh_num == 0xffffffff)
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_BROADCAST_WRITES, 1);
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
} else {
else
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SE_INDEX, se_num);
}
WREG32_SOC15(GC, 0, mmGRBM_GFX_INDEX, data);
}
@ -2425,7 +2365,7 @@ static int gfx_v9_0_kiq_kcq_enable(struct amdgpu_device *adev)
/* 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))) {
if (WARN_ON(i >= (sizeof(queue_mask)*8))) {
DRM_ERROR("Invalid KCQ enabled: %d\n", i);
break;
}
@ -4156,7 +4096,7 @@ static int gfx_v9_0_kiq_irq(struct amdgpu_device *adev,
return 0;
}
const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
static const struct amd_ip_funcs gfx_v9_0_ip_funcs = {
.name = "gfx_v9_0",
.early_init = gfx_v9_0_early_init,
.late_init = gfx_v9_0_late_init,

1
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
View file

@ -24,7 +24,6 @@
#ifndef __GFX_V9_0_H__
#define __GFX_V9_0_H__
extern const struct amd_ip_funcs gfx_v9_0_ip_funcs;
extern const struct amdgpu_ip_block_version gfx_v9_0_ip_block;
void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num);

15
drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c
View file

@ -58,14 +58,14 @@ static void gfxhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
gfxhub_v1_0_init_gart_pt_regs(adev);
WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
(u32)(adev->mc.gtt_start >> 12));
(u32)(adev->mc.gart_start >> 12));
WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
(u32)(adev->mc.gtt_start >> 44));
(u32)(adev->mc.gart_start >> 44));
WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
(u32)(adev->mc.gtt_end >> 12));
(u32)(adev->mc.gart_end >> 12));
WREG32_SOC15(GC, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
(u32)(adev->mc.gtt_end >> 44));
(u32)(adev->mc.gart_end >> 44));
}
static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
@ -129,7 +129,7 @@ static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
/* Setup L2 cache */
tmp = RREG32_SOC15(GC, 0, mmVM_L2_CNTL);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
/* XXX for emulation, Refer to closed source code.*/
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
0);
@ -144,6 +144,8 @@ static void gfxhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
WREG32_SOC15(GC, 0, mmVM_L2_CNTL2, tmp);
tmp = mmVM_L2_CNTL3_DEFAULT;
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
WREG32_SOC15(GC, 0, mmVM_L2_CNTL3, tmp);
tmp = mmVM_L2_CNTL4_DEFAULT;
@ -206,6 +208,9 @@ static void gfxhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
tmp = REG_SET_FIELD(tmp, VM_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, VM_CONTEXT1_CNTL,
RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_CNTL, i, tmp);
WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
WREG32_SOC15_OFFSET(GC, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);

2
drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.h
View file

@ -30,7 +30,5 @@ void gfxhub_v1_0_set_fault_enable_default(struct amdgpu_device *adev,
bool value);
void gfxhub_v1_0_init(struct amdgpu_device *adev);
u64 gfxhub_v1_0_get_mc_fb_offset(struct amdgpu_device *adev);
extern const struct amd_ip_funcs gfxhub_v1_0_ip_funcs;
extern const struct amdgpu_ip_block_version gfxhub_v1_0_ip_block;
#endif

71
drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c
View file

@ -66,14 +66,10 @@ static const u32 crtc_offsets[6] =
SI_CRTC5_REGISTER_OFFSET
};
static void gmc_v6_0_mc_stop(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
static void gmc_v6_0_mc_stop(struct amdgpu_device *adev)
{
u32 blackout;
if (adev->mode_info.num_crtc)
amdgpu_display_stop_mc_access(adev, save);
gmc_v6_0_wait_for_idle((void *)adev);
blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
@ -90,8 +86,7 @@ static void gmc_v6_0_mc_stop(struct amdgpu_device *adev,
}
static void gmc_v6_0_mc_resume(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
static void gmc_v6_0_mc_resume(struct amdgpu_device *adev)
{
u32 tmp;
@ -103,10 +98,6 @@ static void gmc_v6_0_mc_resume(struct amdgpu_device *adev,
tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
WREG32(mmBIF_FB_EN, tmp);
if (adev->mode_info.num_crtc)
amdgpu_display_resume_mc_access(adev, save);
}
static int gmc_v6_0_init_microcode(struct amdgpu_device *adev)
@ -228,20 +219,20 @@ static int gmc_v6_0_mc_load_microcode(struct amdgpu_device *adev)
static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev,
struct amdgpu_mc *mc)
{
u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
base <<= 24;
if (mc->mc_vram_size > 0xFFC0000000ULL) {
dev_warn(adev->dev, "limiting VRAM\n");
mc->real_vram_size = 0xFFC0000000ULL;
mc->mc_vram_size = 0xFFC0000000ULL;
}
amdgpu_vram_location(adev, &adev->mc, 0);
adev->mc.gtt_base_align = 0;
amdgpu_gtt_location(adev, mc);
amdgpu_vram_location(adev, &adev->mc, base);
amdgpu_gart_location(adev, mc);
}
static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
{
struct amdgpu_mode_mc_save save;
u32 tmp;
int i, j;
/* Initialize HDP */
@ -254,16 +245,23 @@ static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
}
WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
if (adev->mode_info.num_crtc)
amdgpu_display_set_vga_render_state(adev, false);
gmc_v6_0_mc_stop(adev, &save);
if (gmc_v6_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for MC idle timedout !\n");
}
WREG32(mmVGA_HDP_CONTROL, VGA_HDP_CONTROL__VGA_MEMORY_DISABLE_MASK);
if (adev->mode_info.num_crtc) {
u32 tmp;
/* Lockout access through VGA aperture*/
tmp = RREG32(mmVGA_HDP_CONTROL);
tmp |= VGA_HDP_CONTROL__VGA_MEMORY_DISABLE_MASK;
WREG32(mmVGA_HDP_CONTROL, tmp);
/* disable VGA render */
tmp = RREG32(mmVGA_RENDER_CONTROL);
tmp &= ~VGA_VSTATUS_CNTL;
WREG32(mmVGA_RENDER_CONTROL, tmp);
}
/* Update configuration */
WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
adev->mc.vram_start >> 12);
@ -271,13 +269,6 @@ static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
adev->mc.vram_end >> 12);
WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
adev->vram_scratch.gpu_addr >> 12);
tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
WREG32(mmMC_VM_FB_LOCATION, tmp);
/* XXX double check these! */
WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
WREG32(mmMC_VM_AGP_BASE, 0);
WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
@ -285,7 +276,6 @@ static void gmc_v6_0_mc_program(struct amdgpu_device *adev)
if (gmc_v6_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for MC idle timedout !\n");
}
gmc_v6_0_mc_resume(adev, &save);
}
static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
@ -342,15 +332,7 @@ static int gmc_v6_0_mc_init(struct amdgpu_device *adev)
adev->mc.real_vram_size = RREG32(mmCONFIG_MEMSIZE) * 1024ULL * 1024ULL;
adev->mc.visible_vram_size = adev->mc.aper_size;
/* unless the user had overridden it, set the gart
* size equal to the 1024 or vram, whichever is larger.
*/
if (amdgpu_gart_size == -1)
adev->mc.gtt_size = max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
adev->mc.mc_vram_size);
else
adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20;
amdgpu_gart_set_defaults(adev);
gmc_v6_0_vram_gtt_location(adev, &adev->mc);
return 0;
@ -511,8 +493,8 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
(4UL << VM_L2_CNTL3__BANK_SELECT__SHIFT) |
(4UL << VM_L2_CNTL3__L2_CACHE_BIGK_FRAGMENT_SIZE__SHIFT));
/* setup context0 */
WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
(u32)(adev->dummy_page.addr >> 12));
@ -559,7 +541,7 @@ static int gmc_v6_0_gart_enable(struct amdgpu_device *adev)
gmc_v6_0_gart_flush_gpu_tlb(adev, 0);
dev_info(adev->dev, "PCIE GART of %uM enabled (table at 0x%016llX).\n",
(unsigned)(adev->mc.gtt_size >> 20),
(unsigned)(adev->mc.gart_size >> 20),
(unsigned long long)adev->gart.table_addr);
adev->gart.ready = true;
return 0;
@ -987,7 +969,6 @@ static int gmc_v6_0_wait_for_idle(void *handle)
static int gmc_v6_0_soft_reset(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct amdgpu_mode_mc_save save;
u32 srbm_soft_reset = 0;
u32 tmp = RREG32(mmSRBM_STATUS);
@ -1003,7 +984,7 @@ static int gmc_v6_0_soft_reset(void *handle)
}
if (srbm_soft_reset) {
gmc_v6_0_mc_stop(adev, &save);
gmc_v6_0_mc_stop(adev);
if (gmc_v6_0_wait_for_idle(adev)) {
dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
}
@ -1023,7 +1004,7 @@ static int gmc_v6_0_soft_reset(void *handle)
udelay(50);
gmc_v6_0_mc_resume(adev, &save);
gmc_v6_0_mc_resume(adev);
udelay(50);
}

68
drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c
View file

@ -37,6 +37,9 @@
#include "oss/oss_2_0_d.h"
#include "oss/oss_2_0_sh_mask.h"
#include "dce/dce_8_0_d.h"
#include "dce/dce_8_0_sh_mask.h"
#include "amdgpu_atombios.h"
static void gmc_v7_0_set_gart_funcs(struct amdgpu_device *adev);
@ -76,14 +79,10 @@ static void gmc_v7_0_init_golden_registers(struct amdgpu_device *adev)
}
}
static void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
static void gmc_v7_0_mc_stop(struct amdgpu_device *adev)
{
u32 blackout;
if (adev->mode_info.num_crtc)
amdgpu_display_stop_mc_access(adev, save);
gmc_v7_0_wait_for_idle((void *)adev);
blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
@ -99,8 +98,7 @@ static void gmc_v7_0_mc_stop(struct amdgpu_device *adev,
udelay(100);
}
static void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
static void gmc_v7_0_mc_resume(struct amdgpu_device *adev)
{
u32 tmp;
@ -112,9 +110,6 @@ static void gmc_v7_0_mc_resume(struct amdgpu_device *adev,
tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
WREG32(mmBIF_FB_EN, tmp);
if (adev->mode_info.num_crtc)
amdgpu_display_resume_mc_access(adev, save);
}
/**
@ -242,15 +237,17 @@ static int gmc_v7_0_mc_load_microcode(struct amdgpu_device *adev)
static void gmc_v7_0_vram_gtt_location(struct amdgpu_device *adev,
struct amdgpu_mc *mc)
{
u64 base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
base <<= 24;
if (mc->mc_vram_size > 0xFFC0000000ULL) {
/* leave room for at least 1024M GTT */
dev_warn(adev->dev, "limiting VRAM\n");
mc->real_vram_size = 0xFFC0000000ULL;
mc->mc_vram_size = 0xFFC0000000ULL;
}
amdgpu_vram_location(adev, &adev->mc, 0);
adev->mc.gtt_base_align = 0;
amdgpu_gtt_location(adev, mc);
amdgpu_vram_location(adev, &adev->mc, base);
amdgpu_gart_location(adev, mc);
}
/**
@ -263,7 +260,6 @@ static void gmc_v7_0_vram_gtt_location(struct amdgpu_device *adev,
*/
static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
{
struct amdgpu_mode_mc_save save;
u32 tmp;
int i, j;
@ -277,13 +273,20 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
}
WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
if (adev->mode_info.num_crtc)
amdgpu_display_set_vga_render_state(adev, false);
gmc_v7_0_mc_stop(adev, &save);
if (gmc_v7_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for MC idle timedout !\n");
}
if (adev->mode_info.num_crtc) {
/* Lockout access through VGA aperture*/
tmp = RREG32(mmVGA_HDP_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
WREG32(mmVGA_HDP_CONTROL, tmp);
/* disable VGA render */
tmp = RREG32(mmVGA_RENDER_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
WREG32(mmVGA_RENDER_CONTROL, tmp);
}
/* Update configuration */
WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
adev->mc.vram_start >> 12);
@ -291,20 +294,12 @@ static void gmc_v7_0_mc_program(struct amdgpu_device *adev)
adev->mc.vram_end >> 12);
WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
adev->vram_scratch.gpu_addr >> 12);
tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
WREG32(mmMC_VM_FB_LOCATION, tmp);
/* XXX double check these! */
WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
WREG32(mmMC_VM_AGP_BASE, 0);
WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
if (gmc_v7_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for MC idle timedout !\n");
}
gmc_v7_0_mc_resume(adev, &save);
WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
@ -391,15 +386,7 @@ static int gmc_v7_0_mc_init(struct amdgpu_device *adev)
if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
adev->mc.visible_vram_size = adev->mc.real_vram_size;
/* unless the user had overridden it, set the gart
* size equal to the 1024 or vram, whichever is larger.
*/
if (amdgpu_gart_size == -1)
adev->mc.gtt_size = max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
adev->mc.mc_vram_size);
else
adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20;
amdgpu_gart_set_defaults(adev);
gmc_v7_0_vram_gtt_location(adev, &adev->mc);
return 0;
@ -611,8 +598,8 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 4);
WREG32(mmVM_L2_CNTL3, tmp);
/* setup context0 */
WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
(u32)(adev->dummy_page.addr >> 12));
@ -666,7 +653,7 @@ static int gmc_v7_0_gart_enable(struct amdgpu_device *adev)
gmc_v7_0_gart_flush_gpu_tlb(adev, 0);
DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
(unsigned)(adev->mc.gtt_size >> 20),
(unsigned)(adev->mc.gart_size >> 20),
(unsigned long long)adev->gart.table_addr);
adev->gart.ready = true;
return 0;
@ -1138,7 +1125,6 @@ static int gmc_v7_0_wait_for_idle(void *handle)
static int gmc_v7_0_soft_reset(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct amdgpu_mode_mc_save save;
u32 srbm_soft_reset = 0;
u32 tmp = RREG32(mmSRBM_STATUS);
@ -1154,7 +1140,7 @@ static int gmc_v7_0_soft_reset(void *handle)
}
if (srbm_soft_reset) {
gmc_v7_0_mc_stop(adev, &save);
gmc_v7_0_mc_stop(adev);
if (gmc_v7_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
}
@ -1175,7 +1161,7 @@ static int gmc_v7_0_soft_reset(void *handle)
/* Wait a little for things to settle down */
udelay(50);
gmc_v7_0_mc_resume(adev, &save);
gmc_v7_0_mc_resume(adev);
udelay(50);
}

81
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
View file

@ -35,6 +35,9 @@
#include "oss/oss_3_0_d.h"
#include "oss/oss_3_0_sh_mask.h"
#include "dce/dce_10_0_d.h"
#include "dce/dce_10_0_sh_mask.h"
#include "vid.h"
#include "vi.h"
@ -161,14 +164,10 @@ static void gmc_v8_0_init_golden_registers(struct amdgpu_device *adev)
}
}
static void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
static void gmc_v8_0_mc_stop(struct amdgpu_device *adev)
{
u32 blackout;
if (adev->mode_info.num_crtc)
amdgpu_display_stop_mc_access(adev, save);
gmc_v8_0_wait_for_idle(adev);
blackout = RREG32(mmMC_SHARED_BLACKOUT_CNTL);
@ -184,8 +183,7 @@ static void gmc_v8_0_mc_stop(struct amdgpu_device *adev,
udelay(100);
}
static void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
struct amdgpu_mode_mc_save *save)
static void gmc_v8_0_mc_resume(struct amdgpu_device *adev)
{
u32 tmp;
@ -197,9 +195,6 @@ static void gmc_v8_0_mc_resume(struct amdgpu_device *adev,
tmp = REG_SET_FIELD(0, BIF_FB_EN, FB_READ_EN, 1);
tmp = REG_SET_FIELD(tmp, BIF_FB_EN, FB_WRITE_EN, 1);
WREG32(mmBIF_FB_EN, tmp);
if (adev->mode_info.num_crtc)
amdgpu_display_resume_mc_access(adev, save);
}
/**
@ -404,15 +399,20 @@ static int gmc_v8_0_polaris_mc_load_microcode(struct amdgpu_device *adev)
static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
struct amdgpu_mc *mc)
{
u64 base = 0;
if (!amdgpu_sriov_vf(adev))
base = RREG32(mmMC_VM_FB_LOCATION) & 0xFFFF;
base <<= 24;
if (mc->mc_vram_size > 0xFFC0000000ULL) {
/* leave room for at least 1024M GTT */
dev_warn(adev->dev, "limiting VRAM\n");
mc->real_vram_size = 0xFFC0000000ULL;
mc->mc_vram_size = 0xFFC0000000ULL;
}
amdgpu_vram_location(adev, &adev->mc, 0);
adev->mc.gtt_base_align = 0;
amdgpu_gtt_location(adev, mc);
amdgpu_vram_location(adev, &adev->mc, base);
amdgpu_gart_location(adev, mc);
}
/**
@ -425,7 +425,6 @@ static void gmc_v8_0_vram_gtt_location(struct amdgpu_device *adev,
*/
static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
{
struct amdgpu_mode_mc_save save;
u32 tmp;
int i, j;
@ -439,13 +438,20 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
}
WREG32(mmHDP_REG_COHERENCY_FLUSH_CNTL, 0);
if (adev->mode_info.num_crtc)
amdgpu_display_set_vga_render_state(adev, false);
gmc_v8_0_mc_stop(adev, &save);
if (gmc_v8_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for MC idle timedout !\n");
}
if (adev->mode_info.num_crtc) {
/* Lockout access through VGA aperture*/
tmp = RREG32(mmVGA_HDP_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
WREG32(mmVGA_HDP_CONTROL, tmp);
/* disable VGA render */
tmp = RREG32(mmVGA_RENDER_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
WREG32(mmVGA_RENDER_CONTROL, tmp);
}
/* Update configuration */
WREG32(mmMC_VM_SYSTEM_APERTURE_LOW_ADDR,
adev->mc.vram_start >> 12);
@ -453,20 +459,23 @@ static void gmc_v8_0_mc_program(struct amdgpu_device *adev)
adev->mc.vram_end >> 12);
WREG32(mmMC_VM_SYSTEM_APERTURE_DEFAULT_ADDR,
adev->vram_scratch.gpu_addr >> 12);
tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
WREG32(mmMC_VM_FB_LOCATION, tmp);
/* XXX double check these! */
WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
if (amdgpu_sriov_vf(adev)) {
tmp = ((adev->mc.vram_end >> 24) & 0xFFFF) << 16;
tmp |= ((adev->mc.vram_start >> 24) & 0xFFFF);
WREG32(mmMC_VM_FB_LOCATION, tmp);
/* XXX double check these! */
WREG32(mmHDP_NONSURFACE_BASE, (adev->mc.vram_start >> 8));
WREG32(mmHDP_NONSURFACE_INFO, (2 << 7) | (1 << 30));
WREG32(mmHDP_NONSURFACE_SIZE, 0x3FFFFFFF);
}
WREG32(mmMC_VM_AGP_BASE, 0);
WREG32(mmMC_VM_AGP_TOP, 0x0FFFFFFF);
WREG32(mmMC_VM_AGP_BOT, 0x0FFFFFFF);
if (gmc_v8_0_wait_for_idle((void *)adev)) {
dev_warn(adev->dev, "Wait for MC idle timedout !\n");
}
gmc_v8_0_mc_resume(adev, &save);
WREG32(mmBIF_FB_EN, BIF_FB_EN__FB_READ_EN_MASK | BIF_FB_EN__FB_WRITE_EN_MASK);
@ -553,15 +562,7 @@ static int gmc_v8_0_mc_init(struct amdgpu_device *adev)
if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
adev->mc.visible_vram_size = adev->mc.real_vram_size;
/* unless the user had overridden it, set the gart
* size equal to the 1024 or vram, whichever is larger.
*/
if (amdgpu_gart_size == -1)
adev->mc.gtt_size = max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
adev->mc.mc_vram_size);
else
adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20;
amdgpu_gart_set_defaults(adev);
gmc_v8_0_vram_gtt_location(adev, &adev->mc);
return 0;
@ -813,8 +814,8 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL4, VMC_TAP_CONTEXT1_PTE_REQUEST_SNOOP, 0);
WREG32(mmVM_L2_CNTL4, tmp);
/* setup context0 */
WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gtt_start >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gtt_end >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_START_ADDR, adev->mc.gart_start >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_END_ADDR, adev->mc.gart_end >> 12);
WREG32(mmVM_CONTEXT0_PAGE_TABLE_BASE_ADDR, adev->gart.table_addr >> 12);
WREG32(mmVM_CONTEXT0_PROTECTION_FAULT_DEFAULT_ADDR,
(u32)(adev->dummy_page.addr >> 12));
@ -869,7 +870,7 @@ static int gmc_v8_0_gart_enable(struct amdgpu_device *adev)
gmc_v8_0_gart_flush_gpu_tlb(adev, 0);
DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
(unsigned)(adev->mc.gtt_size >> 20),
(unsigned)(adev->mc.gart_size >> 20),
(unsigned long long)adev->gart.table_addr);
adev->gart.ready = true;
return 0;
@ -1260,7 +1261,7 @@ static int gmc_v8_0_pre_soft_reset(void *handle)
if (!adev->mc.srbm_soft_reset)
return 0;
gmc_v8_0_mc_stop(adev, &adev->mc.save);
gmc_v8_0_mc_stop(adev);
if (gmc_v8_0_wait_for_idle(adev)) {
dev_warn(adev->dev, "Wait for GMC idle timed out !\n");
}
@ -1306,7 +1307,7 @@ static int gmc_v8_0_post_soft_reset(void *handle)
if (!adev->mc.srbm_soft_reset)
return 0;
gmc_v8_0_mc_resume(adev, &adev->mc.save);
gmc_v8_0_mc_resume(adev);
return 0;
}

137
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
View file

@ -23,11 +23,14 @@
#include <linux/firmware.h>
#include "amdgpu.h"
#include "gmc_v9_0.h"
#include "amdgpu_atomfirmware.h"
#include "vega10/soc15ip.h"
#include "vega10/HDP/hdp_4_0_offset.h"
#include "vega10/HDP/hdp_4_0_sh_mask.h"
#include "vega10/GC/gc_9_0_sh_mask.h"
#include "vega10/DC/dce_12_0_offset.h"
#include "vega10/DC/dce_12_0_sh_mask.h"
#include "vega10/vega10_enum.h"
#include "soc15_common.h"
@ -419,8 +422,7 @@ static void gmc_v9_0_vram_gtt_location(struct amdgpu_device *adev,
if (!amdgpu_sriov_vf(adev))
base = mmhub_v1_0_get_fb_location(adev);
amdgpu_vram_location(adev, &adev->mc, base);
adev->mc.gtt_base_align = 0;
amdgpu_gtt_location(adev, mc);
amdgpu_gart_location(adev, mc);
/* base offset of vram pages */
if (adev->flags & AMD_IS_APU)
adev->vm_manager.vram_base_offset = gfxhub_v1_0_get_mc_fb_offset(adev);
@ -442,43 +444,46 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev)
u32 tmp;
int chansize, numchan;
/* hbm memory channel size */
chansize = 128;
adev->mc.vram_width = amdgpu_atomfirmware_get_vram_width(adev);
if (!adev->mc.vram_width) {
/* hbm memory channel size */
chansize = 128;
tmp = RREG32_SOC15(DF, 0, mmDF_CS_AON0_DramBaseAddress0);
tmp &= DF_CS_AON0_DramBaseAddress0__IntLvNumChan_MASK;
tmp >>= DF_CS_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
switch (tmp) {
case 0:
default:
numchan = 1;
break;
case 1:
numchan = 2;
break;
case 2:
numchan = 0;
break;
case 3:
numchan = 4;
break;
case 4:
numchan = 0;
break;
case 5:
numchan = 8;
break;
case 6:
numchan = 0;
break;
case 7:
numchan = 16;
break;
case 8:
numchan = 2;
break;
tmp = RREG32_SOC15(DF, 0, mmDF_CS_AON0_DramBaseAddress0);
tmp &= DF_CS_AON0_DramBaseAddress0__IntLvNumChan_MASK;
tmp >>= DF_CS_AON0_DramBaseAddress0__IntLvNumChan__SHIFT;
switch (tmp) {
case 0:
default:
numchan = 1;
break;
case 1:
numchan = 2;
break;
case 2:
numchan = 0;
break;
case 3:
numchan = 4;
break;
case 4:
numchan = 0;
break;
case 5:
numchan = 8;
break;
case 6:
numchan = 0;
break;
case 7:
numchan = 16;
break;
case 8:
numchan = 2;
break;
}
adev->mc.vram_width = numchan * chansize;
}
adev->mc.vram_width = numchan * chansize;
/* Could aper size report 0 ? */
adev->mc.aper_base = pci_resource_start(adev->pdev, 0);
@ -494,15 +499,7 @@ static int gmc_v9_0_mc_init(struct amdgpu_device *adev)
if (adev->mc.visible_vram_size > adev->mc.real_vram_size)
adev->mc.visible_vram_size = adev->mc.real_vram_size;
/* unless the user had overridden it, set the gart
* size equal to the 1024 or vram, whichever is larger.
*/
if (amdgpu_gart_size == -1)
adev->mc.gtt_size = max((AMDGPU_DEFAULT_GTT_SIZE_MB << 20),
adev->mc.mc_vram_size);
else
adev->mc.gtt_size = (uint64_t)amdgpu_gart_size << 20;
amdgpu_gart_set_defaults(adev);
gmc_v9_0_vram_gtt_location(adev, &adev->mc);
return 0;
@ -537,10 +534,20 @@ static int gmc_v9_0_sw_init(void *handle)
spin_lock_init(&adev->mc.invalidate_lock);
if (adev->flags & AMD_IS_APU) {
switch (adev->asic_type) {
case CHIP_RAVEN:
adev->mc.vram_type = AMDGPU_VRAM_TYPE_UNKNOWN;
amdgpu_vm_adjust_size(adev, 64);
} else {
if (adev->rev_id == 0x0 || adev->rev_id == 0x1) {
adev->vm_manager.vm_size = 1U << 18;
adev->vm_manager.block_size = 9;
adev->vm_manager.num_level = 3;
} else {
/* vm_size is 64GB for legacy 2-level page support*/
amdgpu_vm_adjust_size(adev, 64);
adev->vm_manager.num_level = 1;
}
break;
case CHIP_VEGA10:
/* XXX Don't know how to get VRAM type yet. */
adev->mc.vram_type = AMDGPU_VRAM_TYPE_HBM;
/*
@ -550,11 +557,16 @@ static int gmc_v9_0_sw_init(void *handle)
*/
adev->vm_manager.vm_size = 1U << 18;
adev->vm_manager.block_size = 9;
DRM_INFO("vm size is %llu GB, block size is %u-bit\n",
adev->vm_manager.vm_size,
adev->vm_manager.block_size);
adev->vm_manager.num_level = 3;
break;
default:
break;
}
DRM_INFO("vm size is %llu GB, block size is %u-bit\n",
adev->vm_manager.vm_size,
adev->vm_manager.block_size);
/* This interrupt is VMC page fault.*/
r = amdgpu_irq_add_id(adev, AMDGPU_IH_CLIENTID_VMC, 0,
&adev->mc.vm_fault);
@ -619,11 +631,6 @@ static int gmc_v9_0_sw_init(void *handle)
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;
/* TODO: fix num_level for APU when updating vm size and block size */
if (adev->flags & AMD_IS_APU)
adev->vm_manager.num_level = 1;
else
adev->vm_manager.num_level = 3;
amdgpu_vm_manager_init(adev);
return 0;
@ -731,7 +738,7 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev)
gmc_v9_0_gart_flush_gpu_tlb(adev, 0);
DRM_INFO("PCIE GART of %uM enabled (table at 0x%016llX).\n",
(unsigned)(adev->mc.gtt_size >> 20),
(unsigned)(adev->mc.gart_size >> 20),
(unsigned long long)adev->gart.table_addr);
adev->gart.ready = true;
return 0;
@ -745,6 +752,20 @@ static int gmc_v9_0_hw_init(void *handle)
/* The sequence of these two function calls matters.*/
gmc_v9_0_init_golden_registers(adev);
if (adev->mode_info.num_crtc) {
u32 tmp;
/* Lockout access through VGA aperture*/
tmp = RREG32_SOC15(DCE, 0, mmVGA_HDP_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_HDP_CONTROL, VGA_MEMORY_DISABLE, 1);
WREG32_SOC15(DCE, 0, mmVGA_HDP_CONTROL, tmp);
/* disable VGA render */
tmp = RREG32_SOC15(DCE, 0, mmVGA_RENDER_CONTROL);
tmp = REG_SET_FIELD(tmp, VGA_RENDER_CONTROL, VGA_VSTATUS_CNTL, 0);
WREG32_SOC15(DCE, 0, mmVGA_RENDER_CONTROL, tmp);
}
r = gmc_v9_0_gart_enable(adev);
return r;

108
drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.c
View file

@ -69,14 +69,14 @@ static void mmhub_v1_0_init_gart_aperture_regs(struct amdgpu_device *adev)
mmhub_v1_0_init_gart_pt_regs(adev);
WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_LO32,
(u32)(adev->mc.gtt_start >> 12));
(u32)(adev->mc.gart_start >> 12));
WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_START_ADDR_HI32,
(u32)(adev->mc.gtt_start >> 44));
(u32)(adev->mc.gart_start >> 44));
WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_LO32,
(u32)(adev->mc.gtt_end >> 12));
(u32)(adev->mc.gart_end >> 12));
WREG32_SOC15(MMHUB, 0, mmVM_CONTEXT0_PAGE_TABLE_END_ADDR_HI32,
(u32)(adev->mc.gtt_end >> 44));
(u32)(adev->mc.gart_end >> 44));
}
static void mmhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev)
@ -143,7 +143,7 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
/* Setup L2 cache */
tmp = RREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_CACHE, 1);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 0);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, ENABLE_L2_FRAGMENT_PROCESSING, 1);
/* XXX for emulation, Refer to closed source code.*/
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL, L2_PDE0_CACHE_TAG_GENERATION_MODE,
0);
@ -158,6 +158,8 @@ static void mmhub_v1_0_init_cache_regs(struct amdgpu_device *adev)
WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL2, tmp);
tmp = mmVM_L2_CNTL3_DEFAULT;
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, BANK_SELECT, 12);
tmp = REG_SET_FIELD(tmp, VM_L2_CNTL3, L2_CACHE_BIGK_FRAGMENT_SIZE, 9);
WREG32_SOC15(MMHUB, 0, mmVM_L2_CNTL3, tmp);
tmp = mmVM_L2_CNTL4_DEFAULT;
@ -222,6 +224,9 @@ static void mmhub_v1_0_setup_vmid_config(struct amdgpu_device *adev)
tmp = REG_SET_FIELD(tmp, VM_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, VM_CONTEXT1_CNTL,
RETRY_PERMISSION_OR_INVALID_PAGE_FAULT, 0);
WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_CNTL, i, tmp);
WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_LO32, i*2, 0);
WREG32_SOC15_OFFSET(MMHUB, 0, mmVM_CONTEXT1_PAGE_TABLE_START_ADDR_HI32, i*2, 0);
@ -245,28 +250,28 @@ static void mmhub_v1_0_program_invalidation(struct amdgpu_device *adev)
}
struct pctl_data {
uint32_t index;
uint32_t data;
uint32_t index;
uint32_t data;
};
const struct pctl_data pctl0_data[] = {
{0x0, 0x7a640},
{0x9, 0x2a64a},
{0xd, 0x2a680},
{0x11, 0x6a684},
{0x19, 0xea68e},
{0x29, 0xa69e},
{0x2b, 0x34a6c0},
{0x61, 0x83a707},
{0xe6, 0x8a7a4},
{0xf0, 0x1a7b8},
{0xf3, 0xfa7cc},
{0x104, 0x17a7dd},
{0x11d, 0xa7dc},
{0x11f, 0x12a7f5},
{0x133, 0xa808},
{0x135, 0x12a810},
{0x149, 0x7a82c}
static const struct pctl_data pctl0_data[] = {
{0x0, 0x7a640},
{0x9, 0x2a64a},
{0xd, 0x2a680},
{0x11, 0x6a684},
{0x19, 0xea68e},
{0x29, 0xa69e},
{0x2b, 0x34a6c0},
{0x61, 0x83a707},
{0xe6, 0x8a7a4},
{0xf0, 0x1a7b8},
{0xf3, 0xfa7cc},
{0x104, 0x17a7dd},
{0x11d, 0xa7dc},
{0x11f, 0x12a7f5},
{0x133, 0xa808},
{0x135, 0x12a810},
{0x149, 0x7a82c}
};
#define PCTL0_DATA_LEN (sizeof(pctl0_data)/sizeof(pctl0_data[0]))
@ -274,32 +279,39 @@ const struct pctl_data pctl0_data[] = {
#define PCTL0_STCTRL_REG_SAVE_RANGE0_BASE 0xa640
#define PCTL0_STCTRL_REG_SAVE_RANGE0_LIMIT 0xa833
const struct pctl_data pctl1_data[] = {
{0x0, 0x39a000},
{0x3b, 0x44a040},
{0x81, 0x2a08d},
{0x85, 0x6ba094},
{0xf2, 0x18a100},
{0x10c, 0x4a132},
{0x112, 0xca141},
{0x120, 0x2fa158},
{0x151, 0x17a1d0},
{0x16a, 0x1a1e9},
{0x16d, 0x13a1ec},
{0x182, 0x7a201},
{0x18b, 0x3a20a},
{0x190, 0x7a580},
{0x199, 0xa590},
{0x19b, 0x4a594},
{0x1a1, 0x1a59c},
{0x1a4, 0x7a82c},
{0x1ad, 0xfa7cc},
{0x1be, 0x17a7dd},
{0x1d7, 0x12a810}
static const struct pctl_data pctl1_data[] = {
{0x0, 0x39a000},
{0x3b, 0x44a040},
{0x81, 0x2a08d},
{0x85, 0x6ba094},
{0xf2, 0x18a100},
{0x10c, 0x4a132},
{0x112, 0xca141},
{0x120, 0x2fa158},
{0x151, 0x17a1d0},
{0x16a, 0x1a1e9},
{0x16d, 0x13a1ec},
{0x182, 0x7a201},
{0x18b, 0x3a20a},
{0x190, 0x7a580},
{0x199, 0xa590},
{0x19b, 0x4a594},
{0x1a1, 0x1a59c},
{0x1a4, 0x7a82c},
{0x1ad, 0xfa7cc},
{0x1be, 0x17a7dd},
{0x1d7, 0x12a810},
{0x1eb, 0x4000a7e1},
{0x1ec, 0x5000a7f5},
{0x1ed, 0x4000a7e2},
{0x1ee, 0x5000a7dc},
{0x1ef, 0x4000a7e3},
{0x1f0, 0x5000a7f6},
{0x1f1, 0x5000a7e4}
};
#define PCTL1_DATA_LEN (sizeof(pctl1_data)/sizeof(pctl1_data[0]))
#define PCTL1_RENG_EXEC_END_PTR 0x1ea
#define PCTL1_RENG_EXEC_END_PTR 0x1f1
#define PCTL1_STCTRL_REG_SAVE_RANGE0_BASE 0xa000
#define PCTL1_STCTRL_REG_SAVE_RANGE0_LIMIT 0xa20d
#define PCTL1_STCTRL_REG_SAVE_RANGE1_BASE 0xa580

3
drivers/gpu/drm/amd/amdgpu/mmhub_v1_0.h
View file

@ -36,7 +36,4 @@ void mmhub_v1_0_initialize_power_gating(struct amdgpu_device *adev);
void mmhub_v1_0_update_power_gating(struct amdgpu_device *adev,
bool enable);
extern const struct amd_ip_funcs mmhub_v1_0_ip_funcs;
extern const struct amdgpu_ip_block_version mmhub_v1_0_ip_block;
#endif

46
drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
View file

@ -72,21 +72,6 @@ static void xgpu_ai_mailbox_set_valid(struct amdgpu_device *adev, bool val)
reg);
}
static void xgpu_ai_mailbox_trans_msg(struct amdgpu_device *adev,
enum idh_request req)
{
u32 reg;
reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0));
reg = REG_SET_FIELD(reg, BIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0,
MSGBUF_DATA, req);
WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0),
reg);
xgpu_ai_mailbox_set_valid(adev, true);
}
static int xgpu_ai_mailbox_rcv_msg(struct amdgpu_device *adev,
enum idh_event event)
{
@ -154,13 +139,25 @@ static int xgpu_ai_poll_msg(struct amdgpu_device *adev, enum idh_event event)
return r;
}
static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
enum idh_request req)
{
static void xgpu_ai_mailbox_trans_msg (struct amdgpu_device *adev,
enum idh_request req, u32 data1, u32 data2, u32 data3) {
u32 reg;
int r;
xgpu_ai_mailbox_trans_msg(adev, req);
reg = RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0));
reg = REG_SET_FIELD(reg, BIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0,
MSGBUF_DATA, req);
WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW0),
reg);
WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW1),
data1);
WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW2),
data2);
WREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0, mmBIF_BX_PF0_MAILBOX_MSGBUF_TRN_DW3),
data3);
xgpu_ai_mailbox_set_valid(adev, true);
/* start to poll ack */
r = xgpu_ai_poll_ack(adev);
@ -168,6 +165,14 @@ static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
pr_err("Doesn't get ack from pf, continue\n");
xgpu_ai_mailbox_set_valid(adev, false);
}
static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
enum idh_request req)
{
int r;
xgpu_ai_mailbox_trans_msg(adev, req, 0, 0, 0);
/* start to check msg if request is idh_req_gpu_init_access */
if (req == IDH_REQ_GPU_INIT_ACCESS ||
@ -342,4 +347,5 @@ const struct amdgpu_virt_ops xgpu_ai_virt_ops = {
.req_full_gpu = xgpu_ai_request_full_gpu_access,
.rel_full_gpu = xgpu_ai_release_full_gpu_access,
.reset_gpu = xgpu_ai_request_reset,
.trans_msg = xgpu_ai_mailbox_trans_msg,
};

4
drivers/gpu/drm/amd/amdgpu/mxgpu_ai.h
View file

@ -31,7 +31,9 @@ enum idh_request {
IDH_REL_GPU_INIT_ACCESS,
IDH_REQ_GPU_FINI_ACCESS,
IDH_REL_GPU_FINI_ACCESS,
IDH_REQ_GPU_RESET_ACCESS
IDH_REQ_GPU_RESET_ACCESS,
IDH_LOG_VF_ERROR = 200,
};
enum idh_event {

1
drivers/gpu/drm/amd/amdgpu/mxgpu_vi.c
View file

@ -613,4 +613,5 @@ const struct amdgpu_virt_ops xgpu_vi_virt_ops = {
.req_full_gpu = xgpu_vi_request_full_gpu_access,
.rel_full_gpu = xgpu_vi_release_full_gpu_access,
.reset_gpu = xgpu_vi_request_reset,
.trans_msg = NULL, /* Does not need to trans VF errors to host. */
};

4
drivers/gpu/drm/amd/amdgpu/mxgpu_vi.h
View file

@ -32,7 +32,9 @@ enum idh_request {
IDH_REL_GPU_INIT_ACCESS,
IDH_REQ_GPU_FINI_ACCESS,
IDH_REL_GPU_FINI_ACCESS,
IDH_REQ_GPU_RESET_ACCESS
IDH_REQ_GPU_RESET_ACCESS,
IDH_LOG_VF_ERROR = 200,
};
/* VI mailbox messages data */

15
drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c
View file

@ -32,6 +32,7 @@
#define smnCPM_CONTROL 0x11180460
#define smnPCIE_CNTL2 0x11180070
#define smnPCIE_CONFIG_CNTL 0x11180044
u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev)
{
@ -67,7 +68,7 @@ void nbio_v6_1_mc_access_enable(struct amdgpu_device *adev, bool enable)
void nbio_v6_1_hdp_flush(struct amdgpu_device *adev)
{
WREG32_SOC15(NBIO, 0, mmBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL, 0);
WREG32_SOC15_NO_KIQ(NBIO, 0, mmBIF_BX_PF0_HDP_MEM_COHERENCY_FLUSH_CNTL, 0);
}
u32 nbio_v6_1_get_memsize(struct amdgpu_device *adev)
@ -256,3 +257,15 @@ void nbio_v6_1_detect_hw_virt(struct amdgpu_device *adev)
adev->virt.caps |= AMDGPU_PASSTHROUGH_MODE;
}
}
void nbio_v6_1_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);
}

1
drivers/gpu/drm/amd/amdgpu/nbio_v6_1.h
View file

@ -50,5 +50,6 @@ void nbio_v6_1_update_medium_grain_clock_gating(struct amdgpu_device *adev, bool
void nbio_v6_1_update_medium_grain_light_sleep(struct amdgpu_device *adev, bool enable);
void nbio_v6_1_get_clockgating_state(struct amdgpu_device *adev, u32 *flags);
void nbio_v6_1_detect_hw_virt(struct amdgpu_device *adev);
void nbio_v6_1_init_registers(struct amdgpu_device *adev);
#endif

2
drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c
View file

@ -65,7 +65,7 @@ void nbio_v7_0_mc_access_enable(struct amdgpu_device *adev, bool enable)
void nbio_v7_0_hdp_flush(struct amdgpu_device *adev)
{
WREG32_SOC15(NBIO, 0, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
WREG32_SOC15_NO_KIQ(NBIO, 0, mmHDP_MEM_COHERENCY_FLUSH_CNTL, 0);
}
u32 nbio_v7_0_get_memsize(struct amdgpu_device *adev)

98
drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
View file

@ -86,6 +86,52 @@ psp_v10_0_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *
return 0;
}
int psp_v10_0_init_microcode(struct psp_context *psp)
{
struct amdgpu_device *adev = psp->adev;
const char *chip_name;
char fw_name[30];
int err = 0;
const struct psp_firmware_header_v1_0 *hdr;
DRM_DEBUG("\n");
switch (adev->asic_type) {
case CHIP_RAVEN:
chip_name = "raven";
break;
default: BUG();
}
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_asd.bin", chip_name);
err = request_firmware(&adev->psp.asd_fw, fw_name, adev->dev);
if (err)
goto out;
err = amdgpu_ucode_validate(adev->psp.asd_fw);
if (err)
goto out;
hdr = (const struct psp_firmware_header_v1_0 *)adev->psp.asd_fw->data;
adev->psp.asd_fw_version = le32_to_cpu(hdr->header.ucode_version);
adev->psp.asd_feature_version = le32_to_cpu(hdr->ucode_feature_version);
adev->psp.asd_ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
adev->psp.asd_start_addr = (uint8_t *)hdr +
le32_to_cpu(hdr->header.ucode_array_offset_bytes);
return 0;
out:
if (err) {
dev_err(adev->dev,
"psp v10.0: Failed to load firmware \"%s\"\n",
fw_name);
release_firmware(adev->psp.asd_fw);
adev->psp.asd_fw = NULL;
}
return err;
}
int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode, struct psp_gfx_cmd_resp *cmd)
{
int ret;
@ -110,7 +156,6 @@ int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode, struct psp_gfx_cm
int psp_v10_0_ring_init(struct psp_context *psp, enum psp_ring_type ring_type)
{
int ret = 0;
unsigned int psp_ring_reg = 0;
struct psp_ring *ring;
struct amdgpu_device *adev = psp->adev;
@ -130,6 +175,16 @@ int psp_v10_0_ring_init(struct psp_context *psp, enum psp_ring_type ring_type)
return ret;
}
return 0;
}
int psp_v10_0_ring_create(struct psp_context *psp, enum psp_ring_type ring_type)
{
int ret = 0;
unsigned int psp_ring_reg = 0;
struct psp_ring *ring = &psp->km_ring;
struct amdgpu_device *adev = psp->adev;
/* Write low address of the ring to C2PMSG_69 */
psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
@ -143,13 +198,42 @@ int psp_v10_0_ring_init(struct psp_context *psp, enum psp_ring_type ring_type)
psp_ring_reg = ring_type;
psp_ring_reg = psp_ring_reg << 16;
WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
/* Wait for response flag (bit 31) in C2PMSG_64 */
psp_ring_reg = 0;
while ((psp_ring_reg & 0x80000000) == 0) {
psp_ring_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64);
}
return 0;
/* There might be handshake issue with hardware which needs delay */
mdelay(20);
/* Wait for response flag (bit 31) in C2PMSG_64 */
ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
0x80000000, 0x8000FFFF, false);
return ret;
}
int psp_v10_0_ring_destroy(struct psp_context *psp, enum psp_ring_type ring_type)
{
int ret = 0;
struct psp_ring *ring;
unsigned int psp_ring_reg = 0;
struct amdgpu_device *adev = psp->adev;
ring = &psp->km_ring;
/* Write the ring destroy command to C2PMSG_64 */
psp_ring_reg = 3 << 16;
WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
/* There might be handshake issue with hardware which needs delay */
mdelay(20);
/* Wait for response flag (bit 31) in C2PMSG_64 */
ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
0x80000000, 0x80000000, false);
amdgpu_bo_free_kernel(&adev->firmware.rbuf,
&ring->ring_mem_mc_addr,
(void **)&ring->ring_mem);
return ret;
}
int psp_v10_0_cmd_submit(struct psp_context *psp,

5
drivers/gpu/drm/amd/amdgpu/psp_v10_0.h
View file

@ -27,10 +27,15 @@
#include "amdgpu_psp.h"
extern int psp_v10_0_init_microcode(struct psp_context *psp);
extern int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd);
extern int psp_v10_0_ring_init(struct psp_context *psp,
enum psp_ring_type ring_type);
extern int psp_v10_0_ring_create(struct psp_context *psp,
enum psp_ring_type ring_type);
extern int psp_v10_0_ring_destroy(struct psp_context *psp,
enum psp_ring_type ring_type);
extern int psp_v10_0_cmd_submit(struct psp_context *psp,
struct amdgpu_firmware_info *ucode,
uint64_t cmd_buf_mc_addr, uint64_t fence_mc_addr,

10
drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
View file

@ -237,11 +237,9 @@ int psp_v3_1_bootloader_load_sos(struct psp_context *psp)
/* there might be handshake issue with hardware which needs delay */
mdelay(20);
#if 0
ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_81),
RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81),
0, true);
#endif
return ret;
}
@ -341,10 +339,10 @@ int psp_v3_1_ring_destroy(struct psp_context *psp, enum psp_ring_type ring_type)
ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
0x80000000, 0x80000000, false);
if (ring->ring_mem)
amdgpu_bo_free_kernel(&adev->firmware.rbuf,
&ring->ring_mem_mc_addr,
(void **)&ring->ring_mem);
amdgpu_bo_free_kernel(&adev->firmware.rbuf,
&ring->ring_mem_mc_addr,
(void **)&ring->ring_mem);
return ret;
}

42
drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
View file

@ -551,17 +551,53 @@ static void sdma_v3_0_rlc_stop(struct amdgpu_device *adev)
*/
static void sdma_v3_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
{
u32 f32_cntl;
u32 f32_cntl, phase_quantum = 0;
int i;
if (amdgpu_sdma_phase_quantum) {
unsigned value = amdgpu_sdma_phase_quantum;
unsigned unit = 0;
while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
value = (value + 1) >> 1;
unit++;
}
if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
WARN_ONCE(1,
"clamping sdma_phase_quantum to %uK clock cycles\n",
value << unit);
}
phase_quantum =
value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
unit << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
}
for (i = 0; i < adev->sdma.num_instances; i++) {
f32_cntl = RREG32(mmSDMA0_CNTL + sdma_offsets[i]);
if (enable)
if (enable) {
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
AUTO_CTXSW_ENABLE, 1);
else
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
ATC_L1_ENABLE, 1);
if (amdgpu_sdma_phase_quantum) {
WREG32(mmSDMA0_PHASE0_QUANTUM + sdma_offsets[i],
phase_quantum);
WREG32(mmSDMA0_PHASE1_QUANTUM + sdma_offsets[i],
phase_quantum);
}
} else {
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
AUTO_CTXSW_ENABLE, 0);
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
ATC_L1_ENABLE, 1);
}
WREG32(mmSDMA0_CNTL + sdma_offsets[i], f32_cntl);
}
}

70
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
View file

@ -291,6 +291,8 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
DRM_DEBUG("Setting write pointer\n");
if (ring->use_doorbell) {
u64 *wb = (u64 *)&adev->wb.wb[ring->wptr_offs];
DRM_DEBUG("Using doorbell -- "
"wptr_offs == 0x%08x "
"lower_32_bits(ring->wptr) << 2 == 0x%08x "
@ -299,8 +301,7 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring)
lower_32_bits(ring->wptr << 2),
upper_32_bits(ring->wptr << 2));
/* XXX check if swapping is necessary on BE */
adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr << 2);
adev->wb.wb[ring->wptr_offs + 1] = upper_32_bits(ring->wptr << 2);
WRITE_ONCE(*wb, (ring->wptr << 2));
DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n",
ring->doorbell_index, ring->wptr << 2);
WDOORBELL64(ring->doorbell_index, ring->wptr << 2);
@ -493,13 +494,45 @@ static void sdma_v4_0_rlc_stop(struct amdgpu_device *adev)
*/
static void sdma_v4_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
{
u32 f32_cntl;
u32 f32_cntl, phase_quantum = 0;
int i;
if (amdgpu_sdma_phase_quantum) {
unsigned value = amdgpu_sdma_phase_quantum;
unsigned unit = 0;
while (value > (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
SDMA0_PHASE0_QUANTUM__VALUE__SHIFT)) {
value = (value + 1) >> 1;
unit++;
}
if (unit > (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
SDMA0_PHASE0_QUANTUM__UNIT__SHIFT)) {
value = (SDMA0_PHASE0_QUANTUM__VALUE_MASK >>
SDMA0_PHASE0_QUANTUM__VALUE__SHIFT);
unit = (SDMA0_PHASE0_QUANTUM__UNIT_MASK >>
SDMA0_PHASE0_QUANTUM__UNIT__SHIFT);
WARN_ONCE(1,
"clamping sdma_phase_quantum to %uK clock cycles\n",
value << unit);
}
phase_quantum =
value << SDMA0_PHASE0_QUANTUM__VALUE__SHIFT |
unit << SDMA0_PHASE0_QUANTUM__UNIT__SHIFT;
}
for (i = 0; i < adev->sdma.num_instances; i++) {
f32_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_CNTL));
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
AUTO_CTXSW_ENABLE, enable ? 1 : 0);
if (enable && amdgpu_sdma_phase_quantum) {
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_PHASE0_QUANTUM),
phase_quantum);
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_PHASE1_QUANTUM),
phase_quantum);
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_PHASE2_QUANTUM),
phase_quantum);
}
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_CNTL), f32_cntl);
}
@ -541,12 +574,13 @@ static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable)
static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
{
struct amdgpu_ring *ring;
u32 rb_cntl, ib_cntl;
u32 rb_cntl, ib_cntl, wptr_poll_cntl;
u32 rb_bufsz;
u32 wb_offset;
u32 doorbell;
u32 doorbell_offset;
u32 temp;
u64 wptr_gpu_addr;
int i, r;
for (i = 0; i < adev->sdma.num_instances; i++) {
@ -628,6 +662,19 @@ static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_F32_CNTL), temp);
}
/* setup the wptr shadow polling */
wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
lower_32_bits(wptr_gpu_addr));
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
upper_32_bits(wptr_gpu_addr));
wptr_poll_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
if (amdgpu_sriov_vf(adev))
wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 1);
else
wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 0);
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), wptr_poll_cntl);
/* enable DMA RB */
rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
@ -655,6 +702,7 @@ static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev)
if (adev->mman.buffer_funcs_ring == ring)
amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size);
}
return 0;
@ -751,15 +799,12 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
const struct sdma_firmware_header_v1_0 *hdr;
const __le32 *fw_data;
u32 fw_size;
u32 digest_size = 0;
int i, j;
/* halt the MEs */
sdma_v4_0_enable(adev, false);
for (i = 0; i < adev->sdma.num_instances; i++) {
uint16_t version_major;
uint16_t version_minor;
if (!adev->sdma.instance[i].fw)
return -EINVAL;
@ -767,23 +812,12 @@ static int sdma_v4_0_load_microcode(struct amdgpu_device *adev)
amdgpu_ucode_print_sdma_hdr(&hdr->header);
fw_size = le32_to_cpu(hdr->header.ucode_size_bytes) / 4;
version_major = le16_to_cpu(hdr->header.header_version_major);
version_minor = le16_to_cpu(hdr->header.header_version_minor);
if (version_major == 1 && version_minor >= 1) {
const struct sdma_firmware_header_v1_1 *sdma_v1_1_hdr = (const struct sdma_firmware_header_v1_1 *) hdr;
digest_size = le32_to_cpu(sdma_v1_1_hdr->digest_size);
}
fw_size -= digest_size;
fw_data = (const __le32 *)
(adev->sdma.instance[i].fw->data +
le32_to_cpu(hdr->header.ucode_array_offset_bytes));
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_ADDR), 0);
for (j = 0; j < fw_size; j++)
WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_UCODE_DATA), le32_to_cpup(fw_data++));

30
drivers/gpu/drm/amd/amdgpu/si.c
View file

@ -1150,6 +1150,33 @@ static bool si_read_disabled_bios(struct amdgpu_device *adev)
return r;
}
#define mmROM_INDEX 0x2A
#define mmROM_DATA 0x2B
static bool si_read_bios_from_rom(struct amdgpu_device *adev,
u8 *bios, u32 length_bytes)
{
u32 *dw_ptr;
u32 i, length_dw;
if (bios == NULL)
return false;
if (length_bytes == 0)
return false;
/* APU vbios image is part of sbios image */
if (adev->flags & AMD_IS_APU)
return false;
dw_ptr = (u32 *)bios;
length_dw = ALIGN(length_bytes, 4) / 4;
/* set rom index to 0 */
WREG32(mmROM_INDEX, 0);
for (i = 0; i < length_dw; i++)
dw_ptr[i] = RREG32(mmROM_DATA);
return true;
}
//xxx: not implemented
static int si_asic_reset(struct amdgpu_device *adev)
{
@ -1206,6 +1233,7 @@ static void si_detect_hw_virtualization(struct amdgpu_device *adev)
static const struct amdgpu_asic_funcs si_asic_funcs =
{
.read_disabled_bios = &si_read_disabled_bios,
.read_bios_from_rom = &si_read_bios_from_rom,
.read_register = &si_read_register,
.reset = &si_asic_reset,
.set_vga_state = &si_vga_set_state,
@ -1385,6 +1413,7 @@ static void si_init_golden_registers(struct amdgpu_device *adev)
amdgpu_program_register_sequence(adev,
pitcairn_mgcg_cgcg_init,
(const u32)ARRAY_SIZE(pitcairn_mgcg_cgcg_init));
break;
case CHIP_VERDE:
amdgpu_program_register_sequence(adev,
verde_golden_registers,
@ -1409,6 +1438,7 @@ static void si_init_golden_registers(struct amdgpu_device *adev)
amdgpu_program_register_sequence(adev,
oland_mgcg_cgcg_init,
(const u32)ARRAY_SIZE(oland_mgcg_cgcg_init));
break;
case CHIP_HAINAN:
amdgpu_program_register_sequence(adev,
hainan_golden_registers,

1
drivers/gpu/drm/amd/amdgpu/si_dpm.c
View file

@ -2055,6 +2055,7 @@ static void si_initialize_powertune_defaults(struct amdgpu_device *adev)
case 0x682C:
si_pi->cac_weights = cac_weights_cape_verde_pro;
si_pi->dte_data = dte_data_sun_xt;
update_dte_from_pl2 = true;
break;
case 0x6825:
case 0x6827:

66
drivers/gpu/drm/amd/amdgpu/soc15.c
View file

@ -25,7 +25,7 @@
#include <linux/module.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_atomfirmware.h"
#include "amdgpu_atombios.h"
#include "amdgpu_ih.h"
#include "amdgpu_uvd.h"
#include "amdgpu_vce.h"
@ -62,8 +62,6 @@
#include "dce_virtual.h"
#include "mxgpu_ai.h"
MODULE_FIRMWARE("amdgpu/vega10_smc.bin");
#define mmFabricConfigAccessControl 0x0410
#define mmFabricConfigAccessControl_BASE_IDX 0
#define mmFabricConfigAccessControl_DEFAULT 0x00000000
@ -198,6 +196,50 @@ static void soc15_didt_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
spin_unlock_irqrestore(&adev->didt_idx_lock, flags);
}
static u32 soc15_gc_cac_rreg(struct amdgpu_device *adev, u32 reg)
{
unsigned long flags;
u32 r;
spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
WREG32_SOC15(GC, 0, mmGC_CAC_IND_INDEX, (reg));
r = RREG32_SOC15(GC, 0, mmGC_CAC_IND_DATA);
spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
return r;
}
static void soc15_gc_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
{
unsigned long flags;
spin_lock_irqsave(&adev->gc_cac_idx_lock, flags);
WREG32_SOC15(GC, 0, mmGC_CAC_IND_INDEX, (reg));
WREG32_SOC15(GC, 0, mmGC_CAC_IND_DATA, (v));
spin_unlock_irqrestore(&adev->gc_cac_idx_lock, flags);
}
static u32 soc15_se_cac_rreg(struct amdgpu_device *adev, u32 reg)
{
unsigned long flags;
u32 r;
spin_lock_irqsave(&adev->se_cac_idx_lock, flags);
WREG32_SOC15(GC, 0, mmSE_CAC_IND_INDEX, (reg));
r = RREG32_SOC15(GC, 0, mmSE_CAC_IND_DATA);
spin_unlock_irqrestore(&adev->se_cac_idx_lock, flags);
return r;
}
static void soc15_se_cac_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
{
unsigned long flags;
spin_lock_irqsave(&adev->se_cac_idx_lock, flags);
WREG32_SOC15(GC, 0, mmSE_CAC_IND_INDEX, (reg));
WREG32_SOC15(GC, 0, mmSE_CAC_IND_DATA, (v));
spin_unlock_irqrestore(&adev->se_cac_idx_lock, flags);
}
static u32 soc15_get_config_memsize(struct amdgpu_device *adev)
{
if (adev->flags & AMD_IS_APU)
@ -392,11 +434,11 @@ static void soc15_gpu_pci_config_reset(struct amdgpu_device *adev)
static int soc15_asic_reset(struct amdgpu_device *adev)
{
amdgpu_atomfirmware_scratch_regs_engine_hung(adev, true);
amdgpu_atombios_scratch_regs_engine_hung(adev, true);
soc15_gpu_pci_config_reset(adev);
amdgpu_atomfirmware_scratch_regs_engine_hung(adev, false);
amdgpu_atombios_scratch_regs_engine_hung(adev, false);
return 0;
}
@ -524,13 +566,6 @@ static uint32_t soc15_get_rev_id(struct amdgpu_device *adev)
return nbio_v6_1_get_rev_id(adev);
}
int gmc_v9_0_mc_wait_for_idle(struct amdgpu_device *adev)
{
/* to be implemented in MC IP*/
return 0;
}
static const struct amdgpu_asic_funcs soc15_asic_funcs =
{
.read_disabled_bios = &soc15_read_disabled_bios,
@ -557,6 +592,10 @@ static int soc15_common_early_init(void *handle)
adev->uvd_ctx_wreg = &soc15_uvd_ctx_wreg;
adev->didt_rreg = &soc15_didt_rreg;
adev->didt_wreg = &soc15_didt_wreg;
adev->gc_cac_rreg = &soc15_gc_cac_rreg;
adev->gc_cac_wreg = &soc15_gc_cac_wreg;
adev->se_cac_rreg = &soc15_se_cac_rreg;
adev->se_cac_wreg = &soc15_se_cac_wreg;
adev->asic_funcs = &soc15_asic_funcs;
@ -681,6 +720,9 @@ static int soc15_common_hw_init(void *handle)
soc15_pcie_gen3_enable(adev);
/* enable aspm */
soc15_program_aspm(adev);
/* setup nbio registers */
if (!(adev->flags & AMD_IS_APU))
nbio_v6_1_init_registers(adev);
/* enable the doorbell aperture */
soc15_enable_doorbell_aperture(adev, true);

7
drivers/gpu/drm/amd/amdgpu/soc15_common.h
View file

@ -77,6 +77,13 @@ struct nbio_pcie_index_data {
(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
(ip##_BASE__INST##inst##_SEG4 + reg))))), value)
#define WREG32_SOC15_NO_KIQ(ip, inst, reg, value) \
WREG32_NO_KIQ( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \
(2 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG2 + reg : \
(3 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG3 + reg : \
(ip##_BASE__INST##inst##_SEG4 + reg))))), value)
#define WREG32_SOC15_OFFSET(ip, inst, reg, offset, value) \
WREG32( (0 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG0 + reg : \
(1 == reg##_BASE_IDX ? ip##_BASE__INST##inst##_SEG1 + reg : \

80
drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
View file

@ -165,6 +165,9 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
unsigned i;
int r;
if (amdgpu_sriov_vf(adev))
return 0;
r = amdgpu_ring_alloc(ring, 16);
if (r) {
DRM_ERROR("amdgpu: uvd enc failed to lock ring %d (%d).\n",
@ -432,13 +435,19 @@ static int uvd_v7_0_sw_init(void *handle)
return r;
}
for (i = 0; i < adev->uvd.num_enc_rings; ++i) {
ring = &adev->uvd.ring_enc[i];
sprintf(ring->name, "uvd_enc%d", i);
if (amdgpu_sriov_vf(adev)) {
ring->use_doorbell = true;
ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
/* currently only use the first enconding ring for
* sriov, so set unused location for other unused rings.
*/
if (i == 0)
ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING0_1 * 2;
else
ring->doorbell_index = AMDGPU_DOORBELL64_UVD_RING2_3 * 2 + 1;
}
r = amdgpu_ring_init(adev, ring, 512, &adev->uvd.irq, 0);
if (r)
@ -685,6 +694,11 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
/* 4, set resp to zero */
WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP, 0);
WDOORBELL32(adev->uvd.ring_enc[0].doorbell_index, 0);
adev->wb.wb[adev->uvd.ring_enc[0].wptr_offs] = 0;
adev->uvd.ring_enc[0].wptr = 0;
adev->uvd.ring_enc[0].wptr_old = 0;
/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
WREG32_SOC15(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST, 0x10000001);
@ -702,7 +716,6 @@ static int uvd_v7_0_mmsch_start(struct amdgpu_device *adev,
dev_err(adev->dev, "failed to init MMSCH, mmVCE_MMSCH_VF_MAILBOX_RESP = %x\n", data);
return -EBUSY;
}
WDOORBELL32(adev->uvd.ring_enc[0].doorbell_index, 0);
return 0;
}
@ -736,11 +749,9 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
init_table += header->uvd_table_offset;
ring = &adev->uvd.ring;
ring->wptr = 0;
size = AMDGPU_GPU_PAGE_ALIGN(adev->uvd.fw->size + 4);
/* disable clock gating */
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS),
~UVD_POWER_STATUS__UVD_PG_MODE_MASK, 0);
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS),
0xFFFFFFFF, 0x00000004);
/* mc resume*/
@ -777,12 +788,6 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CACHE_SIZE2),
AMDGPU_UVD_STACK_SIZE + (AMDGPU_UVD_SESSION_SIZE * 40));
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_UDEC_ADDR_CONFIG),
adev->gfx.config.gb_addr_config);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_UDEC_DB_ADDR_CONFIG),
adev->gfx.config.gb_addr_config);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_UDEC_DBW_ADDR_CONFIG),
adev->gfx.config.gb_addr_config);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_GP_SCRATCH4), adev->uvd.max_handles);
/* mc resume end*/
@ -819,17 +824,6 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
UVD_LMI_CTRL__REQ_MODE_MASK |
0x00100000L));
/* disable byte swapping */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_SWAP_CNTL), 0);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MP_SWAP_CNTL), 0);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXA0), 0x40c2040);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXA1), 0x0);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXB0), 0x40c2040);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUXB1), 0x0);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_ALU), 0);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MPC_SET_MUX), 0x88);
/* take all subblocks out of reset, except VCPU */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET),
UVD_SOFT_RESET__VCPU_SOFT_RESET_MASK);
@ -838,15 +832,6 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_VCPU_CNTL),
UVD_VCPU_CNTL__CLK_EN_MASK);
/* enable UMC */
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
/* boot up the VCPU */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0);
MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0x02, 0x02);
/* enable master interrupt */
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_MASTINT_EN),
~(UVD_MASTINT_EN__VCPU_EN_MASK|UVD_MASTINT_EN__SYS_EN_MASK),
@ -859,40 +844,31 @@ static int uvd_v7_0_sriov_start(struct amdgpu_device *adev)
/* force RBC into idle state */
size = order_base_2(ring->ring_size);
tmp = REG_SET_FIELD(0, UVD_RBC_RB_CNTL, RB_BUFSZ, size);
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_BLKSZ, 1);
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_FETCH, 1);
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_WPTR_POLL_EN, 0);
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_NO_UPDATE, 1);
tmp = REG_SET_FIELD(tmp, UVD_RBC_RB_CNTL, RB_RPTR_WR_EN, 1);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_CNTL), tmp);
/* set the write pointer delay */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_WPTR_CNTL), 0);
/* set the wb address */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RBC_RB_RPTR_ADDR),
(upper_32_bits(ring->gpu_addr) >> 2));
/* programm the RB_BASE for ring buffer */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_LOW),
lower_32_bits(ring->gpu_addr));
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_RBC_RB_64BIT_BAR_HIGH),
upper_32_bits(ring->gpu_addr));
ring->wptr = 0;
ring = &adev->uvd.ring_enc[0];
ring->wptr = 0;
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_LO), ring->gpu_addr);
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_BASE_HI), upper_32_bits(ring->gpu_addr));
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_RB_SIZE), ring->ring_size / 4);
/* boot up the VCPU */
MMSCH_V1_0_INSERT_DIRECT_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_SOFT_RESET), 0);
/* enable UMC */
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(UVD, 0, mmUVD_LMI_CTRL2),
~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK, 0);
MMSCH_V1_0_INSERT_DIRECT_POLL(SOC15_REG_OFFSET(UVD, 0, mmUVD_STATUS), 0x02, 0x02);
/* add end packet */
memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
header->uvd_table_size = table_size;
return uvd_v7_0_mmsch_start(adev, &adev->virt.mm_table);
}
return -EINVAL; /* already initializaed ? */
return uvd_v7_0_mmsch_start(adev, &adev->virt.mm_table);
}
/**

33
drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
View file

@ -173,6 +173,11 @@ static int vce_v4_0_mmsch_start(struct amdgpu_device *adev,
/* 4, set resp to zero */
WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_RESP), 0);
WDOORBELL32(adev->vce.ring[0].doorbell_index, 0);
adev->wb.wb[adev->vce.ring[0].wptr_offs] = 0;
adev->vce.ring[0].wptr = 0;
adev->vce.ring[0].wptr_old = 0;
/* 5, kick off the initialization and wait until VCE_MMSCH_VF_MAILBOX_RESP becomes non-zero */
WREG32(SOC15_REG_OFFSET(VCE, 0, mmVCE_MMSCH_VF_MAILBOX_HOST), 0x10000001);
@ -190,7 +195,6 @@ static int vce_v4_0_mmsch_start(struct amdgpu_device *adev,
dev_err(adev->dev, "failed to init MMSCH, mmVCE_MMSCH_VF_MAILBOX_RESP = %x\n", data);
return -EBUSY;
}
WDOORBELL32(adev->vce.ring[0].doorbell_index, 0);
return 0;
}
@ -274,7 +278,8 @@ static int vce_v4_0_sriov_start(struct amdgpu_device *adev)
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_LMI_CTRL2), ~0x100, 0);
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN),
0xffffffff, VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK,
VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
/* end of MC_RESUME */
MMSCH_V1_0_INSERT_DIRECT_RD_MOD_WT(SOC15_REG_OFFSET(VCE, 0, mmVCE_STATUS),
@ -296,11 +301,9 @@ static int vce_v4_0_sriov_start(struct amdgpu_device *adev)
memcpy((void *)init_table, &end, sizeof(struct mmsch_v1_0_cmd_end));
table_size += sizeof(struct mmsch_v1_0_cmd_end) / 4;
header->vce_table_size = table_size;
return vce_v4_0_mmsch_start(adev, &adev->virt.mm_table);
}
return -EINVAL; /* already initializaed ? */
return vce_v4_0_mmsch_start(adev, &adev->virt.mm_table);
}
/**
@ -443,12 +446,14 @@ static int vce_v4_0_sw_init(void *handle)
if (amdgpu_sriov_vf(adev)) {
/* DOORBELL only works under SRIOV */
ring->use_doorbell = true;
/* currently only use the first encoding ring for sriov,
* so set unused location for other unused rings.
*/
if (i == 0)
ring->doorbell_index = AMDGPU_DOORBELL64_RING0_1 * 2;
else if (i == 1)
ring->doorbell_index = AMDGPU_DOORBELL64_RING2_3 * 2;
ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING0_1 * 2;
else
ring->doorbell_index = AMDGPU_DOORBELL64_RING2_3 * 2 + 1;
ring->doorbell_index = AMDGPU_DOORBELL64_VCE_RING2_3 * 2 + 1;
}
r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0);
if (r)
@ -990,11 +995,13 @@ static int vce_v4_0_set_interrupt_state(struct amdgpu_device *adev,
{
uint32_t val = 0;
if (state == AMDGPU_IRQ_STATE_ENABLE)
val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
if (!amdgpu_sriov_vf(adev)) {
if (state == AMDGPU_IRQ_STATE_ENABLE)
val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK;
WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), val,
~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
WREG32_P(SOC15_REG_OFFSET(VCE, 0, mmVCE_SYS_INT_EN), val,
~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK);
}
return 0;
}

2
drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
View file

@ -664,7 +664,7 @@ static int set_sched_resources(struct device_queue_manager *dqm)
/* This situation may be hit in the future if a new HW
* generation exposes more than 64 queues. If so, the
* definition of res.queue_mask needs updating */
if (WARN_ON(i > (sizeof(res.queue_mask)*8))) {
if (WARN_ON(i >= (sizeof(res.queue_mask)*8))) {
pr_err("Invalid queue enabled by amdgpu: %d\n", i);
break;
}

63
drivers/gpu/drm/amd/include/atomfirmware.h
View file

@ -1233,6 +1233,69 @@ struct atom_asic_profiling_info_v4_1
uint32_t phyclk2gfxclk_c;
};
struct atom_asic_profiling_info_v4_2 {
struct atom_common_table_header table_header;
uint32_t maxvddc;
uint32_t minvddc;
uint32_t avfs_meannsigma_acontant0;
uint32_t avfs_meannsigma_acontant1;
uint32_t avfs_meannsigma_acontant2;
uint16_t avfs_meannsigma_dc_tol_sigma;
uint16_t avfs_meannsigma_platform_mean;
uint16_t avfs_meannsigma_platform_sigma;
uint32_t gb_vdroop_table_cksoff_a0;
uint32_t gb_vdroop_table_cksoff_a1;
uint32_t gb_vdroop_table_cksoff_a2;
uint32_t gb_vdroop_table_ckson_a0;
uint32_t gb_vdroop_table_ckson_a1;
uint32_t gb_vdroop_table_ckson_a2;
uint32_t avfsgb_fuse_table_cksoff_m1;
uint32_t avfsgb_fuse_table_cksoff_m2;
uint32_t avfsgb_fuse_table_cksoff_b;
uint32_t avfsgb_fuse_table_ckson_m1;
uint32_t avfsgb_fuse_table_ckson_m2;
uint32_t avfsgb_fuse_table_ckson_b;
uint16_t max_voltage_0_25mv;
uint8_t enable_gb_vdroop_table_cksoff;
uint8_t enable_gb_vdroop_table_ckson;
uint8_t enable_gb_fuse_table_cksoff;
uint8_t enable_gb_fuse_table_ckson;
uint16_t psm_age_comfactor;
uint8_t enable_apply_avfs_cksoff_voltage;
uint8_t reserved;
uint32_t dispclk2gfxclk_a;
uint32_t dispclk2gfxclk_b;
uint32_t dispclk2gfxclk_c;
uint32_t pixclk2gfxclk_a;
uint32_t pixclk2gfxclk_b;
uint32_t pixclk2gfxclk_c;
uint32_t dcefclk2gfxclk_a;
uint32_t dcefclk2gfxclk_b;
uint32_t dcefclk2gfxclk_c;
uint32_t phyclk2gfxclk_a;
uint32_t phyclk2gfxclk_b;
uint32_t phyclk2gfxclk_c;
uint32_t acg_gb_vdroop_table_a0;
uint32_t acg_gb_vdroop_table_a1;
uint32_t acg_gb_vdroop_table_a2;
uint32_t acg_avfsgb_fuse_table_m1;
uint32_t acg_avfsgb_fuse_table_m2;
uint32_t acg_avfsgb_fuse_table_b;
uint8_t enable_acg_gb_vdroop_table;
uint8_t enable_acg_gb_fuse_table;
uint32_t acg_dispclk2gfxclk_a;
uint32_t acg_dispclk2gfxclk_b;
uint32_t acg_dispclk2gfxclk_c;
uint32_t acg_pixclk2gfxclk_a;
uint32_t acg_pixclk2gfxclk_b;
uint32_t acg_pixclk2gfxclk_c;
uint32_t acg_dcefclk2gfxclk_a;
uint32_t acg_dcefclk2gfxclk_b;
uint32_t acg_dcefclk2gfxclk_c;
uint32_t acg_phyclk2gfxclk_a;
uint32_t acg_phyclk2gfxclk_b;
uint32_t acg_phyclk2gfxclk_c;
};
/*
***************************************************************************

6
drivers/gpu/drm/amd/include/cgs_common.h
View file

@ -50,6 +50,7 @@ enum cgs_ind_reg {
CGS_IND_REG__UVD_CTX,
CGS_IND_REG__DIDT,
CGS_IND_REG_GC_CAC,
CGS_IND_REG_SE_CAC,
CGS_IND_REG__AUDIO_ENDPT
};
@ -406,6 +407,8 @@ typedef int (*cgs_is_virtualization_enabled_t)(void *cgs_device);
typedef int (*cgs_enter_safe_mode)(struct cgs_device *cgs_device, bool en);
typedef void (*cgs_lock_grbm_idx)(struct cgs_device *cgs_device, bool lock);
struct cgs_ops {
/* memory management calls (similar to KFD interface) */
cgs_alloc_gpu_mem_t alloc_gpu_mem;
@ -441,6 +444,7 @@ struct cgs_ops {
cgs_query_system_info query_system_info;
cgs_is_virtualization_enabled_t is_virtualization_enabled;
cgs_enter_safe_mode enter_safe_mode;
cgs_lock_grbm_idx lock_grbm_idx;
};
struct cgs_os_ops; /* To be define in OS-specific CGS header */
@ -517,4 +521,6 @@ struct cgs_device
#define cgs_enter_safe_mode(cgs_device, en) \
CGS_CALL(enter_safe_mode, cgs_device, en)
#define cgs_lock_grbm_idx(cgs_device, lock) \
CGS_CALL(lock_grbm_idx, cgs_device, lock)
#endif /* _CGS_COMMON_H */

113
drivers/gpu/drm/amd/powerplay/hwmgr/cz_hwmgr.c
View file

@ -1240,13 +1240,18 @@ static int cz_phm_force_dpm_highest(struct pp_hwmgr *hwmgr)
{
struct cz_hwmgr *cz_hwmgr = (struct cz_hwmgr *)(hwmgr->backend);
if (cz_hwmgr->sclk_dpm.soft_min_clk !=
cz_hwmgr->sclk_dpm.soft_max_clk)
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMin,
cz_get_sclk_level(hwmgr,
cz_hwmgr->sclk_dpm.soft_max_clk,
PPSMC_MSG_SetSclkSoftMin));
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMin,
cz_get_sclk_level(hwmgr,
cz_hwmgr->sclk_dpm.soft_max_clk,
PPSMC_MSG_SetSclkSoftMin));
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMax,
cz_get_sclk_level(hwmgr,
cz_hwmgr->sclk_dpm.soft_max_clk,
PPSMC_MSG_SetSclkSoftMax));
return 0;
}
@ -1292,17 +1297,55 @@ static int cz_phm_force_dpm_lowest(struct pp_hwmgr *hwmgr)
{
struct cz_hwmgr *cz_hwmgr = (struct cz_hwmgr *)(hwmgr->backend);
if (cz_hwmgr->sclk_dpm.soft_min_clk !=
cz_hwmgr->sclk_dpm.soft_max_clk) {
cz_hwmgr->sclk_dpm.soft_max_clk =
cz_hwmgr->sclk_dpm.soft_min_clk;
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMax,
cz_get_sclk_level(hwmgr,
cz_hwmgr->sclk_dpm.soft_min_clk,
PPSMC_MSG_SetSclkSoftMax));
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMin,
cz_get_sclk_level(hwmgr,
cz_hwmgr->sclk_dpm.soft_min_clk,
PPSMC_MSG_SetSclkSoftMin));
return 0;
}
static int cz_phm_force_dpm_sclk(struct pp_hwmgr *hwmgr, uint32_t sclk)
{
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMin,
cz_get_sclk_level(hwmgr,
sclk,
PPSMC_MSG_SetSclkSoftMin));
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSclkSoftMax,
cz_get_sclk_level(hwmgr,
cz_hwmgr->sclk_dpm.soft_max_clk,
sclk,
PPSMC_MSG_SetSclkSoftMax));
return 0;
}
static int cz_get_profiling_clk(struct pp_hwmgr *hwmgr, uint32_t *sclk)
{
struct phm_clock_voltage_dependency_table *table =
hwmgr->dyn_state.vddc_dependency_on_sclk;
int32_t tmp_sclk;
int32_t count;
tmp_sclk = table->entries[table->count-1].clk * 70 / 100;
for (count = table->count-1; count >= 0; count--) {
if (tmp_sclk >= table->entries[count].clk) {
tmp_sclk = table->entries[count].clk;
*sclk = tmp_sclk;
break;
}
}
if (count < 0)
*sclk = table->entries[0].clk;
return 0;
}
@ -1310,30 +1353,70 @@ static int cz_phm_force_dpm_lowest(struct pp_hwmgr *hwmgr)
static int cz_dpm_force_dpm_level(struct pp_hwmgr *hwmgr,
enum amd_dpm_forced_level level)
{
uint32_t sclk = 0;
int ret = 0;
uint32_t profile_mode_mask = AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD |
AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK |
AMD_DPM_FORCED_LEVEL_PROFILE_PEAK;
if (level == hwmgr->dpm_level)
return ret;
if (!(hwmgr->dpm_level & profile_mode_mask)) {
/* enter profile mode, save current level, disable gfx cg*/
if (level & profile_mode_mask) {
hwmgr->saved_dpm_level = hwmgr->dpm_level;
cgs_set_clockgating_state(hwmgr->device,
AMD_IP_BLOCK_TYPE_GFX,
AMD_CG_STATE_UNGATE);
}
} else {
/* exit profile mode, restore level, enable gfx cg*/
if (!(level & profile_mode_mask)) {
if (level == AMD_DPM_FORCED_LEVEL_PROFILE_EXIT)
level = hwmgr->saved_dpm_level;
cgs_set_clockgating_state(hwmgr->device,
AMD_IP_BLOCK_TYPE_GFX,
AMD_CG_STATE_GATE);
}
}
switch (level) {
case AMD_DPM_FORCED_LEVEL_HIGH:
case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
ret = cz_phm_force_dpm_highest(hwmgr);
if (ret)
return ret;
hwmgr->dpm_level = level;
break;
case AMD_DPM_FORCED_LEVEL_LOW:
case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
ret = cz_phm_force_dpm_lowest(hwmgr);
if (ret)
return ret;
hwmgr->dpm_level = level;
break;
case AMD_DPM_FORCED_LEVEL_AUTO:
ret = cz_phm_unforce_dpm_levels(hwmgr);
if (ret)
return ret;
hwmgr->dpm_level = level;
break;
case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
ret = cz_get_profiling_clk(hwmgr, &sclk);
if (ret)
return ret;
hwmgr->dpm_level = level;
cz_phm_force_dpm_sclk(hwmgr, sclk);
break;
case AMD_DPM_FORCED_LEVEL_MANUAL:
hwmgr->dpm_level = level;
break;
case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT:
default:
break;
}
hwmgr->dpm_level = level;
return ret;
}

25
drivers/gpu/drm/amd/powerplay/hwmgr/hwmgr.c
View file

@ -557,9 +557,8 @@ uint16_t phm_find_closest_vddci(struct pp_atomctrl_voltage_table *vddci_table, u
return vddci_table->entries[i].value;
}
PP_ASSERT_WITH_CODE(false,
"VDDCI is larger than max VDDCI in VDDCI Voltage Table!",
return vddci_table->entries[i-1].value);
pr_debug("vddci is larger than max value in vddci_table\n");
return vddci_table->entries[i-1].value;
}
int phm_find_boot_level(void *table,
@ -583,26 +582,26 @@ int phm_get_sclk_for_voltage_evv(struct pp_hwmgr *hwmgr,
phm_ppt_v1_voltage_lookup_table *lookup_table,
uint16_t virtual_voltage_id, int32_t *sclk)
{
uint8_t entryId;
uint8_t voltageId;
uint8_t entry_id;
uint8_t voltage_id;
struct phm_ppt_v1_information *table_info =
(struct phm_ppt_v1_information *)(hwmgr->pptable);
PP_ASSERT_WITH_CODE(lookup_table->count != 0, "Lookup table is empty", return -EINVAL);
/* search for leakage voltage ID 0xff01 ~ 0xff08 and sckl */
for (entryId = 0; entryId < table_info->vdd_dep_on_sclk->count; entryId++) {
voltageId = table_info->vdd_dep_on_sclk->entries[entryId].vddInd;
if (lookup_table->entries[voltageId].us_vdd == virtual_voltage_id)
for (entry_id = 0; entry_id < table_info->vdd_dep_on_sclk->count; entry_id++) {
voltage_id = table_info->vdd_dep_on_sclk->entries[entry_id].vddInd;
if (lookup_table->entries[voltage_id].us_vdd == virtual_voltage_id)
break;
}
PP_ASSERT_WITH_CODE(entryId < table_info->vdd_dep_on_sclk->count,
"Can't find requested voltage id in vdd_dep_on_sclk table!",
return -EINVAL;
);
if (entry_id >= table_info->vdd_dep_on_sclk->count) {
pr_debug("Can't find requested voltage id in vdd_dep_on_sclk table\n");
return -EINVAL;
}
*sclk = table_info->vdd_dep_on_sclk->entries[entryId].clk;
*sclk = table_info->vdd_dep_on_sclk->entries[entry_id].clk;
return 0;
}

237
drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.c
View file

@ -142,7 +142,7 @@ int pp_atomfwctrl_get_voltage_table_v4(struct pp_hwmgr *hwmgr,
}
} else if (voltage_mode == VOLTAGE_OBJ_SVID2) {
voltage_table->psi1_enable =
voltage_object->svid2_voltage_obj.loadline_psi1 & 0x1;
(voltage_object->svid2_voltage_obj.loadline_psi1 & 0x20) >> 5;
voltage_table->psi0_enable =
voltage_object->svid2_voltage_obj.psi0_enable & 0x1;
voltage_table->max_vid_step =
@ -276,7 +276,10 @@ int pp_atomfwctrl_get_avfs_information(struct pp_hwmgr *hwmgr,
struct pp_atomfwctrl_avfs_parameters *param)
{
uint16_t idx;
uint8_t format_revision, content_revision;
struct atom_asic_profiling_info_v4_1 *profile;
struct atom_asic_profiling_info_v4_2 *profile_v4_2;
idx = GetIndexIntoMasterDataTable(asic_profiling_info);
profile = (struct atom_asic_profiling_info_v4_1 *)
@ -286,76 +289,172 @@ int pp_atomfwctrl_get_avfs_information(struct pp_hwmgr *hwmgr,
if (!profile)
return -1;
param->ulMaxVddc = le32_to_cpu(profile->maxvddc);
param->ulMinVddc = le32_to_cpu(profile->minvddc);
param->ulMeanNsigmaAcontant0 =
le32_to_cpu(profile->avfs_meannsigma_acontant0);
param->ulMeanNsigmaAcontant1 =
le32_to_cpu(profile->avfs_meannsigma_acontant1);
param->ulMeanNsigmaAcontant2 =
le32_to_cpu(profile->avfs_meannsigma_acontant2);
param->usMeanNsigmaDcTolSigma =
le16_to_cpu(profile->avfs_meannsigma_dc_tol_sigma);
param->usMeanNsigmaPlatformMean =
le16_to_cpu(profile->avfs_meannsigma_platform_mean);
param->usMeanNsigmaPlatformSigma =
le16_to_cpu(profile->avfs_meannsigma_platform_sigma);
param->ulGbVdroopTableCksoffA0 =
le32_to_cpu(profile->gb_vdroop_table_cksoff_a0);
param->ulGbVdroopTableCksoffA1 =
le32_to_cpu(profile->gb_vdroop_table_cksoff_a1);
param->ulGbVdroopTableCksoffA2 =
le32_to_cpu(profile->gb_vdroop_table_cksoff_a2);
param->ulGbVdroopTableCksonA0 =
le32_to_cpu(profile->gb_vdroop_table_ckson_a0);
param->ulGbVdroopTableCksonA1 =
le32_to_cpu(profile->gb_vdroop_table_ckson_a1);
param->ulGbVdroopTableCksonA2 =
le32_to_cpu(profile->gb_vdroop_table_ckson_a2);
param->ulGbFuseTableCksoffM1 =
le32_to_cpu(profile->avfsgb_fuse_table_cksoff_m1);
param->ulGbFuseTableCksoffM2 =
le32_to_cpu(profile->avfsgb_fuse_table_cksoff_m2);
param->ulGbFuseTableCksoffB =
le32_to_cpu(profile->avfsgb_fuse_table_cksoff_b);
param->ulGbFuseTableCksonM1 =
le32_to_cpu(profile->avfsgb_fuse_table_ckson_m1);
param->ulGbFuseTableCksonM2 =
le32_to_cpu(profile->avfsgb_fuse_table_ckson_m2);
param->ulGbFuseTableCksonB =
le32_to_cpu(profile->avfsgb_fuse_table_ckson_b);
format_revision = ((struct atom_common_table_header *)profile)->format_revision;
content_revision = ((struct atom_common_table_header *)profile)->content_revision;
param->ucEnableGbVdroopTableCkson =
profile->enable_gb_vdroop_table_ckson;
param->ucEnableGbFuseTableCkson =
profile->enable_gb_fuse_table_ckson;
param->usPsmAgeComfactor =
le16_to_cpu(profile->psm_age_comfactor);
if (format_revision == 4 && content_revision == 1) {
param->ulMaxVddc = le32_to_cpu(profile->maxvddc);
param->ulMinVddc = le32_to_cpu(profile->minvddc);
param->ulMeanNsigmaAcontant0 =
le32_to_cpu(profile->avfs_meannsigma_acontant0);
param->ulMeanNsigmaAcontant1 =
le32_to_cpu(profile->avfs_meannsigma_acontant1);
param->ulMeanNsigmaAcontant2 =
le32_to_cpu(profile->avfs_meannsigma_acontant2);
param->usMeanNsigmaDcTolSigma =
le16_to_cpu(profile->avfs_meannsigma_dc_tol_sigma);
param->usMeanNsigmaPlatformMean =
le16_to_cpu(profile->avfs_meannsigma_platform_mean);
param->usMeanNsigmaPlatformSigma =
le16_to_cpu(profile->avfs_meannsigma_platform_sigma);
param->ulGbVdroopTableCksoffA0 =
le32_to_cpu(profile->gb_vdroop_table_cksoff_a0);
param->ulGbVdroopTableCksoffA1 =
le32_to_cpu(profile->gb_vdroop_table_cksoff_a1);
param->ulGbVdroopTableCksoffA2 =
le32_to_cpu(profile->gb_vdroop_table_cksoff_a2);
param->ulGbVdroopTableCksonA0 =
le32_to_cpu(profile->gb_vdroop_table_ckson_a0);
param->ulGbVdroopTableCksonA1 =
le32_to_cpu(profile->gb_vdroop_table_ckson_a1);
param->ulGbVdroopTableCksonA2 =
le32_to_cpu(profile->gb_vdroop_table_ckson_a2);
param->ulGbFuseTableCksoffM1 =
le32_to_cpu(profile->avfsgb_fuse_table_cksoff_m1);
param->ulGbFuseTableCksoffM2 =
le32_to_cpu(profile->avfsgb_fuse_table_cksoff_m2);
param->ulGbFuseTableCksoffB =
le32_to_cpu(profile->avfsgb_fuse_table_cksoff_b);
param->ulGbFuseTableCksonM1 =
le32_to_cpu(profile->avfsgb_fuse_table_ckson_m1);
param->ulGbFuseTableCksonM2 =
le32_to_cpu(profile->avfsgb_fuse_table_ckson_m2);
param->ulGbFuseTableCksonB =
le32_to_cpu(profile->avfsgb_fuse_table_ckson_b);
param->ulDispclk2GfxclkM1 =
le32_to_cpu(profile->dispclk2gfxclk_a);
param->ulDispclk2GfxclkM2 =
le32_to_cpu(profile->dispclk2gfxclk_b);
param->ulDispclk2GfxclkB =
le32_to_cpu(profile->dispclk2gfxclk_c);
param->ulDcefclk2GfxclkM1 =
le32_to_cpu(profile->dcefclk2gfxclk_a);
param->ulDcefclk2GfxclkM2 =
le32_to_cpu(profile->dcefclk2gfxclk_b);
param->ulDcefclk2GfxclkB =
le32_to_cpu(profile->dcefclk2gfxclk_c);
param->ulPixelclk2GfxclkM1 =
le32_to_cpu(profile->pixclk2gfxclk_a);
param->ulPixelclk2GfxclkM2 =
le32_to_cpu(profile->pixclk2gfxclk_b);
param->ulPixelclk2GfxclkB =
le32_to_cpu(profile->pixclk2gfxclk_c);
param->ulPhyclk2GfxclkM1 =
le32_to_cpu(profile->phyclk2gfxclk_a);
param->ulPhyclk2GfxclkM2 =
le32_to_cpu(profile->phyclk2gfxclk_b);
param->ulPhyclk2GfxclkB =
le32_to_cpu(profile->phyclk2gfxclk_c);
param->ucEnableGbVdroopTableCkson =
profile->enable_gb_vdroop_table_ckson;
param->ucEnableGbFuseTableCkson =
profile->enable_gb_fuse_table_ckson;
param->usPsmAgeComfactor =
le16_to_cpu(profile->psm_age_comfactor);
param->ulDispclk2GfxclkM1 =
le32_to_cpu(profile->dispclk2gfxclk_a);
param->ulDispclk2GfxclkM2 =
le32_to_cpu(profile->dispclk2gfxclk_b);
param->ulDispclk2GfxclkB =
le32_to_cpu(profile->dispclk2gfxclk_c);
param->ulDcefclk2GfxclkM1 =
le32_to_cpu(profile->dcefclk2gfxclk_a);
param->ulDcefclk2GfxclkM2 =
le32_to_cpu(profile->dcefclk2gfxclk_b);
param->ulDcefclk2GfxclkB =
le32_to_cpu(profile->dcefclk2gfxclk_c);
param->ulPixelclk2GfxclkM1 =
le32_to_cpu(profile->pixclk2gfxclk_a);
param->ulPixelclk2GfxclkM2 =
le32_to_cpu(profile->pixclk2gfxclk_b);
param->ulPixelclk2GfxclkB =
le32_to_cpu(profile->pixclk2gfxclk_c);
param->ulPhyclk2GfxclkM1 =
le32_to_cpu(profile->phyclk2gfxclk_a);
param->ulPhyclk2GfxclkM2 =
le32_to_cpu(profile->phyclk2gfxclk_b);
param->ulPhyclk2GfxclkB =
le32_to_cpu(profile->phyclk2gfxclk_c);
param->ulAcgGbVdroopTableA0 = 0;
param->ulAcgGbVdroopTableA1 = 0;
param->ulAcgGbVdroopTableA2 = 0;
param->ulAcgGbFuseTableM1 = 0;
param->ulAcgGbFuseTableM2 = 0;
param->ulAcgGbFuseTableB = 0;
param->ucAcgEnableGbVdroopTable = 0;
param->ucAcgEnableGbFuseTable = 0;
} else if (format_revision == 4 && content_revision == 2) {
profile_v4_2 = (struct atom_asic_profiling_info_v4_2 *)profile;
param->ulMaxVddc = le32_to_cpu(profile_v4_2->maxvddc);
param->ulMinVddc = le32_to_cpu(profile_v4_2->minvddc);
param->ulMeanNsigmaAcontant0 =
le32_to_cpu(profile_v4_2->avfs_meannsigma_acontant0);
param->ulMeanNsigmaAcontant1 =
le32_to_cpu(profile_v4_2->avfs_meannsigma_acontant1);
param->ulMeanNsigmaAcontant2 =
le32_to_cpu(profile_v4_2->avfs_meannsigma_acontant2);
param->usMeanNsigmaDcTolSigma =
le16_to_cpu(profile_v4_2->avfs_meannsigma_dc_tol_sigma);
param->usMeanNsigmaPlatformMean =
le16_to_cpu(profile_v4_2->avfs_meannsigma_platform_mean);
param->usMeanNsigmaPlatformSigma =
le16_to_cpu(profile_v4_2->avfs_meannsigma_platform_sigma);
param->ulGbVdroopTableCksoffA0 =
le32_to_cpu(profile_v4_2->gb_vdroop_table_cksoff_a0);
param->ulGbVdroopTableCksoffA1 =
le32_to_cpu(profile_v4_2->gb_vdroop_table_cksoff_a1);
param->ulGbVdroopTableCksoffA2 =
le32_to_cpu(profile_v4_2->gb_vdroop_table_cksoff_a2);
param->ulGbVdroopTableCksonA0 =
le32_to_cpu(profile_v4_2->gb_vdroop_table_ckson_a0);
param->ulGbVdroopTableCksonA1 =
le32_to_cpu(profile_v4_2->gb_vdroop_table_ckson_a1);
param->ulGbVdroopTableCksonA2 =
le32_to_cpu(profile_v4_2->gb_vdroop_table_ckson_a2);
param->ulGbFuseTableCksoffM1 =
le32_to_cpu(profile_v4_2->avfsgb_fuse_table_cksoff_m1);
param->ulGbFuseTableCksoffM2 =
le32_to_cpu(profile_v4_2->avfsgb_fuse_table_cksoff_m2);
param->ulGbFuseTableCksoffB =
le32_to_cpu(profile_v4_2->avfsgb_fuse_table_cksoff_b);
param->ulGbFuseTableCksonM1 =
le32_to_cpu(profile_v4_2->avfsgb_fuse_table_ckson_m1);
param->ulGbFuseTableCksonM2 =
le32_to_cpu(profile_v4_2->avfsgb_fuse_table_ckson_m2);
param->ulGbFuseTableCksonB =
le32_to_cpu(profile_v4_2->avfsgb_fuse_table_ckson_b);
param->ucEnableGbVdroopTableCkson =
profile_v4_2->enable_gb_vdroop_table_ckson;
param->ucEnableGbFuseTableCkson =
profile_v4_2->enable_gb_fuse_table_ckson;
param->usPsmAgeComfactor =
le16_to_cpu(profile_v4_2->psm_age_comfactor);
param->ulDispclk2GfxclkM1 =
le32_to_cpu(profile_v4_2->dispclk2gfxclk_a);
param->ulDispclk2GfxclkM2 =
le32_to_cpu(profile_v4_2->dispclk2gfxclk_b);
param->ulDispclk2GfxclkB =
le32_to_cpu(profile_v4_2->dispclk2gfxclk_c);
param->ulDcefclk2GfxclkM1 =
le32_to_cpu(profile_v4_2->dcefclk2gfxclk_a);
param->ulDcefclk2GfxclkM2 =
le32_to_cpu(profile_v4_2->dcefclk2gfxclk_b);
param->ulDcefclk2GfxclkB =
le32_to_cpu(profile_v4_2->dcefclk2gfxclk_c);
param->ulPixelclk2GfxclkM1 =
le32_to_cpu(profile_v4_2->pixclk2gfxclk_a);
param->ulPixelclk2GfxclkM2 =
le32_to_cpu(profile_v4_2->pixclk2gfxclk_b);
param->ulPixelclk2GfxclkB =
le32_to_cpu(profile_v4_2->pixclk2gfxclk_c);
param->ulPhyclk2GfxclkM1 =
le32_to_cpu(profile->phyclk2gfxclk_a);
param->ulPhyclk2GfxclkM2 =
le32_to_cpu(profile_v4_2->phyclk2gfxclk_b);
param->ulPhyclk2GfxclkB =
le32_to_cpu(profile_v4_2->phyclk2gfxclk_c);
param->ulAcgGbVdroopTableA0 = le32_to_cpu(profile_v4_2->acg_gb_vdroop_table_a0);
param->ulAcgGbVdroopTableA1 = le32_to_cpu(profile_v4_2->acg_gb_vdroop_table_a1);
param->ulAcgGbVdroopTableA2 = le32_to_cpu(profile_v4_2->acg_gb_vdroop_table_a2);
param->ulAcgGbFuseTableM1 = le32_to_cpu(profile_v4_2->acg_avfsgb_fuse_table_m1);
param->ulAcgGbFuseTableM2 = le32_to_cpu(profile_v4_2->acg_avfsgb_fuse_table_m2);
param->ulAcgGbFuseTableB = le32_to_cpu(profile_v4_2->acg_avfsgb_fuse_table_b);
param->ucAcgEnableGbVdroopTable = le32_to_cpu(profile_v4_2->enable_acg_gb_vdroop_table);
param->ucAcgEnableGbFuseTable = le32_to_cpu(profile_v4_2->enable_acg_gb_fuse_table);
} else {
pr_info("Invalid VBIOS AVFS ProfilingInfo Revision!\n");
return -EINVAL;
}
return 0;
}

8
drivers/gpu/drm/amd/powerplay/hwmgr/ppatomfwctrl.h
View file

@ -109,6 +109,14 @@ struct pp_atomfwctrl_avfs_parameters {
uint32_t ulPhyclk2GfxclkM1;
uint32_t ulPhyclk2GfxclkM2;
uint32_t ulPhyclk2GfxclkB;
uint32_t ulAcgGbVdroopTableA0;
uint32_t ulAcgGbVdroopTableA1;
uint32_t ulAcgGbVdroopTableA2;
uint32_t ulAcgGbFuseTableM1;
uint32_t ulAcgGbFuseTableM2;
uint32_t ulAcgGbFuseTableB;
uint32_t ucAcgEnableGbVdroopTable;
uint32_t ucAcgEnableGbFuseTable;
};
struct pp_atomfwctrl_gpio_parameters {

9
drivers/gpu/drm/amd/powerplay/hwmgr/rv_hwmgr.c
View file

@ -265,6 +265,15 @@ static int rv_tf_set_clock_limit(struct pp_hwmgr *hwmgr, void *input,
}
} */
if (((hwmgr->uvd_arbiter.vclk_soft_min / 100) != rv_data->vclk_soft_min) ||
((hwmgr->uvd_arbiter.dclk_soft_min / 100) != rv_data->dclk_soft_min)) {
rv_data->vclk_soft_min = hwmgr->uvd_arbiter.vclk_soft_min / 100;
rv_data->dclk_soft_min = hwmgr->uvd_arbiter.dclk_soft_min / 100;
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,
PPSMC_MSG_SetSoftMinVcn,
(rv_data->vclk_soft_min << 16) | rv_data->vclk_soft_min);
}
if((hwmgr->gfx_arbiter.sclk_hard_min != 0) &&
((hwmgr->gfx_arbiter.sclk_hard_min / 100) != rv_data->soc_actual_hard_min_freq)) {
smum_send_msg_to_smc_with_parameter(hwmgr->smumgr,

2
drivers/gpu/drm/amd/powerplay/hwmgr/rv_hwmgr.h
View file

@ -280,6 +280,8 @@ struct rv_hwmgr {
uint32_t f_actual_hard_min_freq;
uint32_t fabric_actual_soft_min_freq;
uint32_t vclk_soft_min;
uint32_t dclk_soft_min;
uint32_t gfx_actual_soft_min_freq;
bool vcn_power_gated;

Some files were not shown because too many files have changed in this diff Show more