UbiquitousOS
/
OpenCloudOS-Kernel
mirror of https://gitee.com/OpenCloudOS/OpenCloudOS-Kernel.git
11
0
Fork 0
Code Issues Projects Releases Wiki Activity

Merge tag 'amd-drm-next-5.7-2020-03-19' of git://people.freedesktop.org/~agd5f/linux into drm-next 

amd-drm-next-5.7-2020-03-19:

amdgpu:
- SR-IOV fixes
- RAS fixes
- Fallthrough cleanups
- Kconfig fix for ACP
- Fix load balancing with VCN
- DC fixes
- GPU reset fixes
- Various cleanups

scheduler:
- Revert job distribution optimization
- Add a helper to pick the least loaded scheduler

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200319175418.4237-1-alexander.deucher@amd.com
This commit is contained in:
Dave Airlie 2020-03-20 12:54:51 +10:00
parent 7c2cb99fe7 8cd296082c
commit 042539396a
88 changed files with 960 additions and 508 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
drivers/gpu/drm/amd/acp/Kconfig
View File

@ -1,5 +1,6 @@
# SPDX-License-Identifier: MIT
menu "ACP (Audio CoProcessor) Configuration"
depends on DRM_AMDGPU
config DRM_AMD_ACP
bool "Enable AMD Audio CoProcessor IP support"

2
drivers/gpu/drm/amd/amdgpu/amdgpu.h
View File

@ -994,6 +994,8 @@ uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
uint32_t acc_flags);
void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
uint32_t acc_flags);
void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
uint32_t acc_flags);
void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value);
uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset);

2
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
View File

@ -79,7 +79,7 @@ static uint32_t get_sdma_rlc_reg_offset(struct amdgpu_device *adev,
dev_warn(adev->dev,
"Invalid sdma engine id (%d), using engine id 0\n",
engine_id);
/* fall through */
fallthrough;
case 0:
sdma_engine_reg_base = SOC15_REG_OFFSET(SDMA0, 0,
mmSDMA0_RLC0_RB_CNTL) - mmSDMA0_RLC0_RB_CNTL;

86
drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
View File

@ -91,47 +91,51 @@ static int amdgpu_ctx_init_entity(struct amdgpu_ctx *ctx, const u32 hw_ip, const
priority = (ctx->override_priority == DRM_SCHED_PRIORITY_UNSET) ?
ctx->init_priority : ctx->override_priority;
switch (hw_ip) {
case AMDGPU_HW_IP_GFX:
sched = &adev->gfx.gfx_ring[0].sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_COMPUTE:
hw_prio = amdgpu_ctx_sched_prio_to_compute_prio(priority);
scheds = adev->gfx.compute_prio_sched[hw_prio];
num_scheds = adev->gfx.num_compute_sched[hw_prio];
break;
case AMDGPU_HW_IP_DMA:
scheds = adev->sdma.sdma_sched;
num_scheds = adev->sdma.num_sdma_sched;
break;
case AMDGPU_HW_IP_UVD:
sched = &adev->uvd.inst[0].ring.sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_VCE:
sched = &adev->vce.ring[0].sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_UVD_ENC:
sched = &adev->uvd.inst[0].ring_enc[0].sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_VCN_DEC:
scheds = adev->vcn.vcn_dec_sched;
num_scheds = adev->vcn.num_vcn_dec_sched;
break;
case AMDGPU_HW_IP_VCN_ENC:
scheds = adev->vcn.vcn_enc_sched;
num_scheds = adev->vcn.num_vcn_enc_sched;
break;
case AMDGPU_HW_IP_VCN_JPEG:
scheds = adev->jpeg.jpeg_sched;
num_scheds = adev->jpeg.num_jpeg_sched;
break;
case AMDGPU_HW_IP_GFX:
sched = &adev->gfx.gfx_ring[0].sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_COMPUTE:
hw_prio = amdgpu_ctx_sched_prio_to_compute_prio(priority);
scheds = adev->gfx.compute_prio_sched[hw_prio];
num_scheds = adev->gfx.num_compute_sched[hw_prio];
break;
case AMDGPU_HW_IP_DMA:
scheds = adev->sdma.sdma_sched;
num_scheds = adev->sdma.num_sdma_sched;
break;
case AMDGPU_HW_IP_UVD:
sched = &adev->uvd.inst[0].ring.sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_VCE:
sched = &adev->vce.ring[0].sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_UVD_ENC:
sched = &adev->uvd.inst[0].ring_enc[0].sched;
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_VCN_DEC:
sched = drm_sched_pick_best(adev->vcn.vcn_dec_sched,
adev->vcn.num_vcn_dec_sched);
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_VCN_ENC:
sched = drm_sched_pick_best(adev->vcn.vcn_enc_sched,
adev->vcn.num_vcn_enc_sched);
scheds = &sched;
num_scheds = 1;
break;
case AMDGPU_HW_IP_VCN_JPEG:
scheds = adev->jpeg.jpeg_sched;
num_scheds = adev->jpeg.num_jpeg_sched;
break;
}
r = drm_sched_entity_init(&entity->entity, priority, scheds, num_scheds,

10
drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
View File

@ -33,6 +33,7 @@
#include "amdgpu.h"
#include "amdgpu_pm.h"
#include "amdgpu_dm_debugfs.h"
#include "amdgpu_ras.h"
/**
* amdgpu_debugfs_add_files - Add simple debugfs entries
@ -178,7 +179,7 @@ static int amdgpu_debugfs_process_reg_op(bool read, struct file *f,
} else {
r = get_user(value, (uint32_t *)buf);
if (!r)
WREG32(*pos >> 2, value);
amdgpu_mm_wreg_mmio_rlc(adev, *pos >> 2, value, 0);
}
if (r) {
result = r;
@ -783,11 +784,11 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
ssize_t result = 0;
uint32_t offset, se, sh, cu, wave, simd, thread, bank, *data;
if (size & 3 || *pos & 3)
if (size > 4096 || size & 3 || *pos & 3)
return -EINVAL;
/* decode offset */
offset = *pos & GENMASK_ULL(11, 0);
offset = (*pos & GENMASK_ULL(11, 0)) >> 2;
se = (*pos & GENMASK_ULL(19, 12)) >> 12;
sh = (*pos & GENMASK_ULL(27, 20)) >> 20;
cu = (*pos & GENMASK_ULL(35, 28)) >> 28;
@ -825,7 +826,7 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
while (size) {
uint32_t value;
value = data[offset++];
value = data[result >> 2];
r = put_user(value, (uint32_t *)buf);
if (r) {
result = r;
@ -1294,7 +1295,6 @@ DEFINE_SIMPLE_ATTRIBUTE(fops_ib_preempt, NULL,
DEFINE_SIMPLE_ATTRIBUTE(fops_sclk_set, NULL,
amdgpu_debugfs_sclk_set, "%llu\n");
extern void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev);
int amdgpu_debugfs_init(struct amdgpu_device *adev)
{
int r, i;

66
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
View File

@ -306,28 +306,10 @@ void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value)
BUG();
}
/**
* amdgpu_mm_wreg - write to a memory mapped IO register
*
* @adev: amdgpu_device pointer
* @reg: dword aligned register offset
* @v: 32 bit value to write to the register
* @acc_flags: access flags which require special behavior
*
* Writes the value specified to the offset specified.
*/
void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
uint32_t acc_flags)
void static inline amdgpu_mm_wreg_mmio(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint32_t acc_flags)
{
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_KIQ) || (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev)))
return amdgpu_kiq_wreg(adev, reg, v);
if ((reg * 4) < adev->rmmio_size && !(acc_flags & AMDGPU_REGS_IDX))
writel(v, ((void __iomem *)adev->rmmio) + (reg * 4));
else {
@ -344,6 +326,48 @@ void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
}
}
/**
* amdgpu_mm_wreg - write to a memory mapped IO register
*
* @adev: amdgpu_device pointer
* @reg: dword aligned register offset
* @v: 32 bit value to write to the register
* @acc_flags: access flags which require special behavior
*
* Writes the value specified to the offset specified.
*/
void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
uint32_t acc_flags)
{
if (adev->asic_type >= CHIP_VEGA10 && reg == 0) {
adev->last_mm_index = v;
}
if ((acc_flags & AMDGPU_REGS_KIQ) || (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev)))
return amdgpu_kiq_wreg(adev, reg, v);
amdgpu_mm_wreg_mmio(adev, reg, v, acc_flags);
}
/*
* amdgpu_mm_wreg_mmio_rlc - write register either with mmio or with RLC path if in range
*
* this function is invoked only the debugfs register access
* */
void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
uint32_t acc_flags)
{
if (amdgpu_sriov_fullaccess(adev) &&
adev->gfx.rlc.funcs &&
adev->gfx.rlc.funcs->is_rlcg_access_range) {
if (adev->gfx.rlc.funcs->is_rlcg_access_range(adev, reg))
return adev->gfx.rlc.funcs->rlcg_wreg(adev, reg, v);
}
amdgpu_mm_wreg_mmio(adev, reg, v, acc_flags);
}
/**
* amdgpu_io_rreg - read an IO register
*
@ -3933,6 +3957,8 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive,
if (r)
goto out;
amdgpu_fbdev_set_suspend(tmp_adev, 0);
/* must succeed. */
amdgpu_ras_resume(tmp_adev);
@ -4106,6 +4132,8 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
*/
amdgpu_unregister_gpu_instance(tmp_adev);
amdgpu_fbdev_set_suspend(adev, 1);
/* disable ras on ALL IPs */
if (!(in_ras_intr && !use_baco) &&
amdgpu_device_ip_need_full_reset(tmp_adev))

2
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
View File

@ -448,6 +448,8 @@ struct amdgpu_pm {
/* powerplay feature */
uint32_t pp_feature;
/* Used for I2C access to various EEPROMs on relevant ASICs */
struct i2c_adapter smu_i2c;
};
#define R600_SSTU_DFLT 0

15
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
View File

@ -159,6 +159,10 @@ static int psp_sw_fini(void *handle)
adev->psp.sos_fw = NULL;
release_firmware(adev->psp.asd_fw);
adev->psp.asd_fw = NULL;
if (adev->psp.cap_fw) {
release_firmware(adev->psp.cap_fw);
adev->psp.cap_fw = NULL;
}
if (adev->psp.ta_fw) {
release_firmware(adev->psp.ta_fw);
adev->psp.ta_fw = NULL;
@ -200,6 +204,7 @@ psp_cmd_submit_buf(struct psp_context *psp,
int ret;
int index;
int timeout = 2000;
bool ras_intr = false;
mutex_lock(&psp->mutex);
@ -224,7 +229,8 @@ psp_cmd_submit_buf(struct psp_context *psp,
* because gpu reset thread triggered and lock resource should
* be released for psp resume sequence.
*/
if (amdgpu_ras_intr_triggered())
ras_intr = amdgpu_ras_intr_triggered();
if (ras_intr)
break;
msleep(1);
amdgpu_asic_invalidate_hdp(psp->adev, NULL);
@ -237,14 +243,14 @@ psp_cmd_submit_buf(struct psp_context *psp,
* during psp initialization to avoid breaking hw_init and it doesn't
* return -EINVAL.
*/
if (psp->cmd_buf_mem->resp.status || !timeout) {
if ((psp->cmd_buf_mem->resp.status || !timeout) && !ras_intr) {
if (ucode)
DRM_WARN("failed to load ucode id (%d) ",
ucode->ucode_id);
DRM_WARN("psp command (0x%X) failed and response status is (0x%X)\n",
psp->cmd_buf_mem->cmd_id,
psp->cmd_buf_mem->resp.status);
if (!timeout) {
if ((ucode->ucode_id == AMDGPU_UCODE_ID_CAP) || !timeout) {
mutex_unlock(&psp->mutex);
return -EINVAL;
}
@ -1186,6 +1192,9 @@ static int psp_get_fw_type(struct amdgpu_firmware_info *ucode,
enum psp_gfx_fw_type *type)
{
switch (ucode->ucode_id) {
case AMDGPU_UCODE_ID_CAP:
*type = GFX_FW_TYPE_CAP;
break;
case AMDGPU_UCODE_ID_SDMA0:
*type = GFX_FW_TYPE_SDMA0;
break;

3
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
View File

@ -252,6 +252,9 @@ struct psp_context
uint32_t asd_ucode_size;
uint8_t *asd_start_addr;
/* cap firmware */
const struct firmware *cap_fw;
/* fence buffer */
struct amdgpu_bo *fence_buf_bo;
uint64_t fence_buf_mc_addr;

31
drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
View File

@ -1116,7 +1116,7 @@ void amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
{
struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
struct ras_manager *obj, *tmp;
struct ras_manager *obj;
struct ras_fs_if fs_info;
/*
@ -1128,10 +1128,7 @@ void amdgpu_ras_debugfs_create_all(struct amdgpu_device *adev)
amdgpu_ras_debugfs_create_ctrl_node(adev);
list_for_each_entry_safe(obj, tmp, &con->head, node) {
if (!obj)
continue;
list_for_each_entry(obj, &con->head, node) {
if (amdgpu_ras_is_supported(adev, obj->head.block) &&
(obj->attr_inuse == 1)) {
sprintf(fs_info.debugfs_name, "%s_err_inject",
@ -1765,18 +1762,30 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev,
*hw_supported = 0;
*supported = 0;
if (amdgpu_sriov_vf(adev) ||
if (amdgpu_sriov_vf(adev) || !adev->is_atom_fw ||
(adev->asic_type != CHIP_VEGA20 &&
adev->asic_type != CHIP_ARCTURUS))
return;
if (adev->is_atom_fw &&
(amdgpu_atomfirmware_mem_ecc_supported(adev) ||
amdgpu_atomfirmware_sram_ecc_supported(adev)))
*hw_supported = AMDGPU_RAS_BLOCK_MASK;
if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
DRM_INFO("HBM ECC is active.\n");
*hw_supported |= (1 << AMDGPU_RAS_BLOCK__UMC |
1 << AMDGPU_RAS_BLOCK__DF);
} else
DRM_INFO("HBM ECC is not presented.\n");
if (amdgpu_atomfirmware_sram_ecc_supported(adev)) {
DRM_INFO("SRAM ECC is active.\n");
*hw_supported |= ~(1 << AMDGPU_RAS_BLOCK__UMC |
1 << AMDGPU_RAS_BLOCK__DF);
} else
DRM_INFO("SRAM ECC is not presented.\n");
/* hw_supported needs to be aligned with RAS block mask. */
*hw_supported &= AMDGPU_RAS_BLOCK_MASK;
*supported = amdgpu_ras_enable == 0 ?
0 : *hw_supported & amdgpu_ras_mask;
0 : *hw_supported & amdgpu_ras_mask;
}
int amdgpu_ras_init(struct amdgpu_device *adev)

48
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
View File

@ -25,7 +25,6 @@
#include "amdgpu.h"
#include "amdgpu_ras.h"
#include <linux/bits.h>
#include "smu_v11_0_i2c.h"
#include "atom.h"
#define EEPROM_I2C_TARGET_ADDR_VEGA20 0xA0
@ -124,6 +123,7 @@ static int __update_table_header(struct amdgpu_ras_eeprom_control *control,
unsigned char *buff)
{
int ret = 0;
struct amdgpu_device *adev = to_amdgpu_device(control);
struct i2c_msg msg = {
.addr = 0,
.flags = 0,
@ -137,7 +137,7 @@ static int __update_table_header(struct amdgpu_ras_eeprom_control *control,
msg.addr = control->i2c_address;
ret = i2c_transfer(&control->eeprom_accessor, &msg, 1);
ret = i2c_transfer(&adev->pm.smu_i2c, &msg, 1);
if (ret < 1)
DRM_ERROR("Failed to write EEPROM table header, ret:%d", ret);
@ -251,33 +251,18 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
.buf = buff,
};
/* Verify i2c adapter is initialized */
if (!adev->pm.smu_i2c.algo)
return -ENOENT;
if (!__get_eeprom_i2c_addr(adev, &control->i2c_address))
return -EINVAL;
mutex_init(&control->tbl_mutex);
switch (adev->asic_type) {
case CHIP_VEGA20:
ret = smu_v11_0_i2c_eeprom_control_init(&control->eeprom_accessor);
break;
case CHIP_ARCTURUS:
ret = smu_i2c_eeprom_init(&adev->smu, &control->eeprom_accessor);
break;
default:
return 0;
}
if (ret) {
DRM_ERROR("Failed to init I2C controller, ret:%d", ret);
return ret;
}
msg.addr = control->i2c_address;
/* Read/Create table header from EEPROM address 0 */
ret = i2c_transfer(&control->eeprom_accessor, &msg, 1);
ret = i2c_transfer(&adev->pm.smu_i2c, &msg, 1);
if (ret < 1) {
DRM_ERROR("Failed to read EEPROM table header, ret:%d", ret);
return ret;
@ -303,23 +288,6 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control)
return ret == 1 ? 0 : -EIO;
}
void amdgpu_ras_eeprom_fini(struct amdgpu_ras_eeprom_control *control)
{
struct amdgpu_device *adev = to_amdgpu_device(control);
switch (adev->asic_type) {
case CHIP_VEGA20:
smu_v11_0_i2c_eeprom_control_fini(&control->eeprom_accessor);
break;
case CHIP_ARCTURUS:
smu_i2c_eeprom_fini(&adev->smu, &control->eeprom_accessor);
break;
default:
return;
}
}
static void __encode_table_record_to_buff(struct amdgpu_ras_eeprom_control *control,
struct eeprom_table_record *record,
unsigned char *buff)
@ -476,7 +444,7 @@ int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control,
control->next_addr += EEPROM_TABLE_RECORD_SIZE;
}
ret = i2c_transfer(&control->eeprom_accessor, msgs, num);
ret = i2c_transfer(&adev->pm.smu_i2c, msgs, num);
if (ret < 1) {
DRM_ERROR("Failed to process EEPROM table records, ret:%d", ret);

2
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h
View File

@ -44,7 +44,6 @@ struct amdgpu_ras_eeprom_table_header {
struct amdgpu_ras_eeprom_control {
struct amdgpu_ras_eeprom_table_header tbl_hdr;
struct i2c_adapter eeprom_accessor;
uint32_t next_addr;
unsigned int num_recs;
struct mutex tbl_mutex;
@ -79,7 +78,6 @@ struct eeprom_table_record {
}__attribute__((__packed__));
int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control);
void amdgpu_ras_eeprom_fini(struct amdgpu_ras_eeprom_control *control);
int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control);
int amdgpu_ras_eeprom_process_recods(struct amdgpu_ras_eeprom_control *control,

2
drivers/gpu/drm/amd/amdgpu/amdgpu_rlc.h
View File

@ -127,6 +127,8 @@ struct amdgpu_rlc_funcs {
void (*reset)(struct amdgpu_device *adev);
void (*start)(struct amdgpu_device *adev);
void (*update_spm_vmid)(struct amdgpu_device *adev, unsigned vmid);
void (*rlcg_wreg)(struct amdgpu_device *adev, u32 offset, u32 v);
bool (*is_rlcg_access_range)(struct amdgpu_device *adev, uint32_t reg);
};
struct amdgpu_rlc {

3
drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
View File

@ -283,7 +283,8 @@ union amdgpu_firmware_header {
* fw loading support
*/
enum AMDGPU_UCODE_ID {
AMDGPU_UCODE_ID_SDMA0 = 0,
AMDGPU_UCODE_ID_CAP = 0, /* CAP must be the 1st fw to be loaded */
AMDGPU_UCODE_ID_SDMA0,
AMDGPU_UCODE_ID_SDMA1,
AMDGPU_UCODE_ID_SDMA2,
AMDGPU_UCODE_ID_SDMA3,

3
drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
View File

@ -270,6 +270,9 @@ struct amdgpu_virt {
#define amdgpu_sriov_runtime(adev) \
((adev)->virt.caps & AMDGPU_SRIOV_CAPS_RUNTIME)
#define amdgpu_sriov_fullaccess(adev) \
(amdgpu_sriov_vf((adev)) && !amdgpu_sriov_runtime((adev)))
#define amdgpu_passthrough(adev) \
((adev)->virt.caps & AMDGPU_PASSTHROUGH_MODE)

22
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
View File

@ -1446,7 +1446,7 @@ static int amdgpu_vm_update_ptes(struct amdgpu_vm_update_params *params,
uint64_t incr, entry_end, pe_start;
struct amdgpu_bo *pt;
if (flags & AMDGPU_PTE_VALID) {
if (flags & (AMDGPU_PTE_VALID | AMDGPU_PTE_PRT)) {
/* make sure that the page tables covering the
* address range are actually allocated
*/
@ -1603,14 +1603,11 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
goto error_unlock;
}
if (flags & AMDGPU_PTE_VALID) {
if (flags & (AMDGPU_PTE_VALID | AMDGPU_PTE_PRT)) {
struct amdgpu_bo *root = vm->root.base.bo;
if (!dma_fence_is_signaled(vm->last_direct))
amdgpu_bo_fence(root, vm->last_direct, true);
if (!dma_fence_is_signaled(vm->last_delayed))
amdgpu_bo_fence(root, vm->last_delayed, true);
}
r = vm->update_funcs->prepare(&params, resv, sync_mode);
@ -1718,7 +1715,7 @@ static int amdgpu_vm_bo_split_mapping(struct amdgpu_device *adev,
AMDGPU_GPU_PAGES_IN_CPU_PAGE;
}
} else if (flags & AMDGPU_PTE_VALID) {
} else if (flags & (AMDGPU_PTE_VALID | AMDGPU_PTE_PRT)) {
addr += bo_adev->vm_manager.vram_base_offset;
addr += pfn << PAGE_SHIFT;
}
@ -2588,8 +2585,7 @@ bool amdgpu_vm_evictable(struct amdgpu_bo *bo)
return false;
/* Don't evict VM page tables while they are updated */
if (!dma_fence_is_signaled(bo_base->vm->last_direct) ||
!dma_fence_is_signaled(bo_base->vm->last_delayed)) {
if (!dma_fence_is_signaled(bo_base->vm->last_direct)) {
amdgpu_vm_eviction_unlock(bo_base->vm);
return false;
}
@ -2766,11 +2762,7 @@ long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout)
if (timeout <= 0)
return timeout;
timeout = dma_fence_wait_timeout(vm->last_direct, true, timeout);
if (timeout <= 0)
return timeout;
return dma_fence_wait_timeout(vm->last_delayed, true, timeout);
return dma_fence_wait_timeout(vm->last_direct, true, timeout);
}
/**
@ -2843,7 +2835,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
vm->update_funcs = &amdgpu_vm_sdma_funcs;
vm->last_update = NULL;
vm->last_direct = dma_fence_get_stub();
vm->last_delayed = dma_fence_get_stub();
mutex_init(&vm->eviction_lock);
vm->evicting = false;
@ -2898,7 +2889,6 @@ error_free_root:
error_free_delayed:
dma_fence_put(vm->last_direct);
dma_fence_put(vm->last_delayed);
drm_sched_entity_destroy(&vm->delayed);
error_free_direct:
@ -3101,8 +3091,6 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
dma_fence_wait(vm->last_direct, false);
dma_fence_put(vm->last_direct);
dma_fence_wait(vm->last_delayed, false);
dma_fence_put(vm->last_delayed);
list_for_each_entry_safe(mapping, tmp, &vm->freed, list) {
if (mapping->flags & AMDGPU_PTE_PRT && prt_fini_needed) {

1
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
View File

@ -276,7 +276,6 @@ struct amdgpu_vm {
/* Last submission to the scheduler entities */
struct dma_fence *last_direct;
struct dma_fence *last_delayed;
unsigned int pasid;
/* dedicated to vm */

11
drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
View File

@ -104,12 +104,13 @@ static int amdgpu_vm_sdma_commit(struct amdgpu_vm_update_params *p,
if (r)
goto error;
tmp = dma_fence_get(f);
if (p->direct)
if (p->direct) {
tmp = dma_fence_get(f);
swap(p->vm->last_direct, tmp);
else
swap(p->vm->last_delayed, tmp);
dma_fence_put(tmp);
dma_fence_put(tmp);
} else {
dma_resv_add_shared_fence(p->vm->root.base.bo->tbo.base.resv, f);
}
if (fence && !p->direct)
swap(*fence, f);

74
drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
View File

@ -224,6 +224,49 @@ static const struct soc15_reg_golden golden_settings_gc_10_1_2[] =
SOC15_REG_GOLDEN_VALUE(GC, 0, mmUTCL1_CTRL, 0xffffffff, 0x00800000)
};
static void gfx_v10_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 v)
{
static void *scratch_reg0;
static void *scratch_reg1;
static void *scratch_reg2;
static void *scratch_reg3;
static void *spare_int;
static uint32_t grbm_cntl;
static uint32_t grbm_idx;
uint32_t i = 0;
uint32_t retries = 50000;
scratch_reg0 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0)*4;
scratch_reg1 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1)*4;
scratch_reg2 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG2)*4;
scratch_reg3 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3)*4;
spare_int = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT)*4;
grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL;
grbm_idx = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_INDEX_BASE_IDX] + mmGRBM_GFX_INDEX;
if (amdgpu_sriov_runtime(adev)) {
pr_err("shouldn't call rlcg write register during runtime\n");
return;
}
writel(v, scratch_reg0);
writel(offset | 0x80000000, scratch_reg1);
writel(1, spare_int);
for (i = 0; i < retries; i++) {
u32 tmp;
tmp = readl(scratch_reg1);
if (!(tmp & 0x80000000))
break;
udelay(10);
}
if (i >= retries)
pr_err("timeout: rlcg program reg:0x%05x failed !\n", offset);
}
static const struct soc15_reg_golden golden_settings_gc_10_1_nv14[] =
{
/* Pending on emulation bring up */
@ -4247,6 +4290,33 @@ static void gfx_v10_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
WREG32_SOC15(GC, 0, mmRLC_SPM_MC_CNTL, data);
}
static bool gfx_v10_0_check_rlcg_range(struct amdgpu_device *adev,
uint32_t offset,
struct soc15_reg_rlcg *entries, int arr_size)
{
int i;
uint32_t reg;
if (!entries)
return false;
for (i = 0; i < arr_size; i++) {
const struct soc15_reg_rlcg *entry;
entry = &entries[i];
reg = adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
if (offset == reg)
return true;
}
return false;
}
static bool gfx_v10_0_is_rlcg_access_range(struct amdgpu_device *adev, u32 offset)
{
return gfx_v10_0_check_rlcg_range(adev, offset, NULL, 0);
}
static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs = {
.is_rlc_enabled = gfx_v10_0_is_rlc_enabled,
.set_safe_mode = gfx_v10_0_set_safe_mode,
@ -4258,7 +4328,9 @@ static const struct amdgpu_rlc_funcs gfx_v10_0_rlc_funcs = {
.stop = gfx_v10_0_rlc_stop,
.reset = gfx_v10_0_rlc_reset,
.start = gfx_v10_0_rlc_start,
.update_spm_vmid = gfx_v10_0_update_spm_vmid
.update_spm_vmid = gfx_v10_0_update_spm_vmid,
.rlcg_wreg = gfx_v10_rlcg_wreg,
.is_rlcg_access_range = gfx_v10_0_is_rlcg_access_range,
};
static int gfx_v10_0_set_powergating_state(void *handle,

237
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
View File

@ -697,6 +697,11 @@ static const struct soc15_reg_golden golden_settings_gc_9_4_1_arct[] =
SOC15_REG_GOLDEN_VALUE(GC, 0, mmSQ_FIFO_SIZES, 0xffffffff, 0x00000f00),
};
static const struct soc15_reg_rlcg rlcg_access_gc_9_0[] = {
{SOC15_REG_ENTRY(GC, 0, mmGRBM_GFX_INDEX)},
{SOC15_REG_ENTRY(GC, 0, mmSQ_IND_INDEX)},
};
static const u32 GFX_RLC_SRM_INDEX_CNTL_ADDR_OFFSETS[] =
{
mmRLC_SRM_INDEX_CNTL_ADDR_0 - mmRLC_SRM_INDEX_CNTL_ADDR_0,
@ -721,6 +726,59 @@ static const u32 GFX_RLC_SRM_INDEX_CNTL_DATA_OFFSETS[] =
mmRLC_SRM_INDEX_CNTL_DATA_7 - mmRLC_SRM_INDEX_CNTL_DATA_0,
};
void gfx_v9_0_rlcg_wreg(struct amdgpu_device *adev, u32 offset, u32 v)
{
static void *scratch_reg0;
static void *scratch_reg1;
static void *scratch_reg2;
static void *scratch_reg3;
static void *spare_int;
static uint32_t grbm_cntl;
static uint32_t grbm_idx;
scratch_reg0 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0)*4;
scratch_reg1 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG1)*4;
scratch_reg2 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG2)*4;
scratch_reg3 = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3)*4;
spare_int = adev->rmmio + (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_BASE_IDX] + mmRLC_SPARE_INT)*4;
grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL;
grbm_idx = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_INDEX_BASE_IDX] + mmGRBM_GFX_INDEX;
if (amdgpu_sriov_runtime(adev)) {
pr_err("shouldn't call rlcg write register during runtime\n");
return;
}
if (offset == grbm_cntl || offset == grbm_idx) {
if (offset == grbm_cntl)
writel(v, scratch_reg2);
else if (offset == grbm_idx)
writel(v, scratch_reg3);
writel(v, ((void __iomem *)adev->rmmio) + (offset * 4));
} else {
uint32_t i = 0;
uint32_t retries = 50000;
writel(v, scratch_reg0);
writel(offset | 0x80000000, scratch_reg1);
writel(1, spare_int);
for (i = 0; i < retries; i++) {
u32 tmp;
tmp = readl(scratch_reg1);
if (!(tmp & 0x80000000))
break;
udelay(10);
}
if (i >= retries)
pr_err("timeout: rlcg program reg:0x%05x failed !\n", offset);
}
}
#define VEGA10_GB_ADDR_CONFIG_GOLDEN 0x2a114042
#define VEGA12_GB_ADDR_CONFIG_GOLDEN 0x24104041
#define RAVEN_GB_ADDR_CONFIG_GOLDEN 0x24000042
@ -1921,7 +1979,7 @@ static int gfx_v9_0_mec_init(struct amdgpu_device *adev)
static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
{
WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
WREG32_SOC15_RLC(GC, 0, mmSQ_IND_INDEX,
(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
(address << SQ_IND_INDEX__INDEX__SHIFT) |
@ -1933,7 +1991,7 @@ static void wave_read_regs(struct amdgpu_device *adev, uint32_t simd,
uint32_t wave, uint32_t thread,
uint32_t regno, uint32_t num, uint32_t *out)
{
WREG32_SOC15(GC, 0, mmSQ_IND_INDEX,
WREG32_SOC15_RLC(GC, 0, mmSQ_IND_INDEX,
(wave << SQ_IND_INDEX__WAVE_ID__SHIFT) |
(simd << SQ_IND_INDEX__SIMD_ID__SHIFT) |
(regno << SQ_IND_INDEX__INDEX__SHIFT) |
@ -4128,6 +4186,101 @@ static const u32 sgpr_init_compute_shader[] =
0xbe800080, 0xbf810000,
};
static const u32 vgpr_init_compute_shader_arcturus[] = {
0xd3d94000, 0x18000080, 0xd3d94001, 0x18000080, 0xd3d94002, 0x18000080,
0xd3d94003, 0x18000080, 0xd3d94004, 0x18000080, 0xd3d94005, 0x18000080,
0xd3d94006, 0x18000080, 0xd3d94007, 0x18000080, 0xd3d94008, 0x18000080,
0xd3d94009, 0x18000080, 0xd3d9400a, 0x18000080, 0xd3d9400b, 0x18000080,
0xd3d9400c, 0x18000080, 0xd3d9400d, 0x18000080, 0xd3d9400e, 0x18000080,
0xd3d9400f, 0x18000080, 0xd3d94010, 0x18000080, 0xd3d94011, 0x18000080,
0xd3d94012, 0x18000080, 0xd3d94013, 0x18000080, 0xd3d94014, 0x18000080,
0xd3d94015, 0x18000080, 0xd3d94016, 0x18000080, 0xd3d94017, 0x18000080,
0xd3d94018, 0x18000080, 0xd3d94019, 0x18000080, 0xd3d9401a, 0x18000080,
0xd3d9401b, 0x18000080, 0xd3d9401c, 0x18000080, 0xd3d9401d, 0x18000080,
0xd3d9401e, 0x18000080, 0xd3d9401f, 0x18000080, 0xd3d94020, 0x18000080,
0xd3d94021, 0x18000080, 0xd3d94022, 0x18000080, 0xd3d94023, 0x18000080,
0xd3d94024, 0x18000080, 0xd3d94025, 0x18000080, 0xd3d94026, 0x18000080,
0xd3d94027, 0x18000080, 0xd3d94028, 0x18000080, 0xd3d94029, 0x18000080,
0xd3d9402a, 0x18000080, 0xd3d9402b, 0x18000080, 0xd3d9402c, 0x18000080,
0xd3d9402d, 0x18000080, 0xd3d9402e, 0x18000080, 0xd3d9402f, 0x18000080,
0xd3d94030, 0x18000080, 0xd3d94031, 0x18000080, 0xd3d94032, 0x18000080,
0xd3d94033, 0x18000080, 0xd3d94034, 0x18000080, 0xd3d94035, 0x18000080,
0xd3d94036, 0x18000080, 0xd3d94037, 0x18000080, 0xd3d94038, 0x18000080,
0xd3d94039, 0x18000080, 0xd3d9403a, 0x18000080, 0xd3d9403b, 0x18000080,
0xd3d9403c, 0x18000080, 0xd3d9403d, 0x18000080, 0xd3d9403e, 0x18000080,
0xd3d9403f, 0x18000080, 0xd3d94040, 0x18000080, 0xd3d94041, 0x18000080,
0xd3d94042, 0x18000080, 0xd3d94043, 0x18000080, 0xd3d94044, 0x18000080,
0xd3d94045, 0x18000080, 0xd3d94046, 0x18000080, 0xd3d94047, 0x18000080,
0xd3d94048, 0x18000080, 0xd3d94049, 0x18000080, 0xd3d9404a, 0x18000080,
0xd3d9404b, 0x18000080, 0xd3d9404c, 0x18000080, 0xd3d9404d, 0x18000080,
0xd3d9404e, 0x18000080, 0xd3d9404f, 0x18000080, 0xd3d94050, 0x18000080,
0xd3d94051, 0x18000080, 0xd3d94052, 0x18000080, 0xd3d94053, 0x18000080,
0xd3d94054, 0x18000080, 0xd3d94055, 0x18000080, 0xd3d94056, 0x18000080,
0xd3d94057, 0x18000080, 0xd3d94058, 0x18000080, 0xd3d94059, 0x18000080,
0xd3d9405a, 0x18000080, 0xd3d9405b, 0x18000080, 0xd3d9405c, 0x18000080,
0xd3d9405d, 0x18000080, 0xd3d9405e, 0x18000080, 0xd3d9405f, 0x18000080,
0xd3d94060, 0x18000080, 0xd3d94061, 0x18000080, 0xd3d94062, 0x18000080,
0xd3d94063, 0x18000080, 0xd3d94064, 0x18000080, 0xd3d94065, 0x18000080,
0xd3d94066, 0x18000080, 0xd3d94067, 0x18000080, 0xd3d94068, 0x18000080,
0xd3d94069, 0x18000080, 0xd3d9406a, 0x18000080, 0xd3d9406b, 0x18000080,
0xd3d9406c, 0x18000080, 0xd3d9406d, 0x18000080, 0xd3d9406e, 0x18000080,
0xd3d9406f, 0x18000080, 0xd3d94070, 0x18000080, 0xd3d94071, 0x18000080,
0xd3d94072, 0x18000080, 0xd3d94073, 0x18000080, 0xd3d94074, 0x18000080,
0xd3d94075, 0x18000080, 0xd3d94076, 0x18000080, 0xd3d94077, 0x18000080,
0xd3d94078, 0x18000080, 0xd3d94079, 0x18000080, 0xd3d9407a, 0x18000080,
0xd3d9407b, 0x18000080, 0xd3d9407c, 0x18000080, 0xd3d9407d, 0x18000080,
0xd3d9407e, 0x18000080, 0xd3d9407f, 0x18000080, 0xd3d94080, 0x18000080,
0xd3d94081, 0x18000080, 0xd3d94082, 0x18000080, 0xd3d94083, 0x18000080,
0xd3d94084, 0x18000080, 0xd3d94085, 0x18000080, 0xd3d94086, 0x18000080,
0xd3d94087, 0x18000080, 0xd3d94088, 0x18000080, 0xd3d94089, 0x18000080,
0xd3d9408a, 0x18000080, 0xd3d9408b, 0x18000080, 0xd3d9408c, 0x18000080,
0xd3d9408d, 0x18000080, 0xd3d9408e, 0x18000080, 0xd3d9408f, 0x18000080,
0xd3d94090, 0x18000080, 0xd3d94091, 0x18000080, 0xd3d94092, 0x18000080,
0xd3d94093, 0x18000080, 0xd3d94094, 0x18000080, 0xd3d94095, 0x18000080,
0xd3d94096, 0x18000080, 0xd3d94097, 0x18000080, 0xd3d94098, 0x18000080,
0xd3d94099, 0x18000080, 0xd3d9409a, 0x18000080, 0xd3d9409b, 0x18000080,
0xd3d9409c, 0x18000080, 0xd3d9409d, 0x18000080, 0xd3d9409e, 0x18000080,
0xd3d9409f, 0x18000080, 0xd3d940a0, 0x18000080, 0xd3d940a1, 0x18000080,
0xd3d940a2, 0x18000080, 0xd3d940a3, 0x18000080, 0xd3d940a4, 0x18000080,
0xd3d940a5, 0x18000080, 0xd3d940a6, 0x18000080, 0xd3d940a7, 0x18000080,
0xd3d940a8, 0x18000080, 0xd3d940a9, 0x18000080, 0xd3d940aa, 0x18000080,
0xd3d940ab, 0x18000080, 0xd3d940ac, 0x18000080, 0xd3d940ad, 0x18000080,
0xd3d940ae, 0x18000080, 0xd3d940af, 0x18000080, 0xd3d940b0, 0x18000080,
0xd3d940b1, 0x18000080, 0xd3d940b2, 0x18000080, 0xd3d940b3, 0x18000080,
0xd3d940b4, 0x18000080, 0xd3d940b5, 0x18000080, 0xd3d940b6, 0x18000080,
0xd3d940b7, 0x18000080, 0xd3d940b8, 0x18000080, 0xd3d940b9, 0x18000080,
0xd3d940ba, 0x18000080, 0xd3d940bb, 0x18000080, 0xd3d940bc, 0x18000080,
0xd3d940bd, 0x18000080, 0xd3d940be, 0x18000080, 0xd3d940bf, 0x18000080,
0xd3d940c0, 0x18000080, 0xd3d940c1, 0x18000080, 0xd3d940c2, 0x18000080,
0xd3d940c3, 0x18000080, 0xd3d940c4, 0x18000080, 0xd3d940c5, 0x18000080,
0xd3d940c6, 0x18000080, 0xd3d940c7, 0x18000080, 0xd3d940c8, 0x18000080,
0xd3d940c9, 0x18000080, 0xd3d940ca, 0x18000080, 0xd3d940cb, 0x18000080,
0xd3d940cc, 0x18000080, 0xd3d940cd, 0x18000080, 0xd3d940ce, 0x18000080,
0xd3d940cf, 0x18000080, 0xd3d940d0, 0x18000080, 0xd3d940d1, 0x18000080,
0xd3d940d2, 0x18000080, 0xd3d940d3, 0x18000080, 0xd3d940d4, 0x18000080,
0xd3d940d5, 0x18000080, 0xd3d940d6, 0x18000080, 0xd3d940d7, 0x18000080,
0xd3d940d8, 0x18000080, 0xd3d940d9, 0x18000080, 0xd3d940da, 0x18000080,
0xd3d940db, 0x18000080, 0xd3d940dc, 0x18000080, 0xd3d940dd, 0x18000080,
0xd3d940de, 0x18000080, 0xd3d940df, 0x18000080, 0xd3d940e0, 0x18000080,
0xd3d940e1, 0x18000080, 0xd3d940e2, 0x18000080, 0xd3d940e3, 0x18000080,
0xd3d940e4, 0x18000080, 0xd3d940e5, 0x18000080, 0xd3d940e6, 0x18000080,
0xd3d940e7, 0x18000080, 0xd3d940e8, 0x18000080, 0xd3d940e9, 0x18000080,
0xd3d940ea, 0x18000080, 0xd3d940eb, 0x18000080, 0xd3d940ec, 0x18000080,
0xd3d940ed, 0x18000080, 0xd3d940ee, 0x18000080, 0xd3d940ef, 0x18000080,
0xd3d940f0, 0x18000080, 0xd3d940f1, 0x18000080, 0xd3d940f2, 0x18000080,
0xd3d940f3, 0x18000080, 0xd3d940f4, 0x18000080, 0xd3d940f5, 0x18000080,
0xd3d940f6, 0x18000080, 0xd3d940f7, 0x18000080, 0xd3d940f8, 0x18000080,
0xd3d940f9, 0x18000080, 0xd3d940fa, 0x18000080, 0xd3d940fb, 0x18000080,
0xd3d940fc, 0x18000080, 0xd3d940fd, 0x18000080, 0xd3d940fe, 0x18000080,
0xd3d940ff, 0x18000080, 0xb07c0000, 0xbe8a00ff, 0x000000f8, 0xbf11080a,
0x7e000280, 0x7e020280, 0x7e040280, 0x7e060280, 0x7e080280, 0x7e0a0280,
0x7e0c0280, 0x7e0e0280, 0x808a880a, 0xbe80320a, 0xbf84fff5, 0xbf9c0000,
0xd28c0001, 0x0001007f, 0xd28d0001, 0x0002027e, 0x10020288, 0xb88b0904,
0xb78b4000, 0xd1196a01, 0x00001701, 0xbe8a0087, 0xbefc00c1, 0xd89c4000,
0x00020201, 0xd89cc080, 0x00040401, 0x320202ff, 0x00000800, 0x808a810a,
0xbf84fff8, 0xbf810000,
};
/* When below register arrays changed, please update gpr_reg_size,
and sec_ded_counter_reg_size in function gfx_v9_0_do_edc_gpr_workarounds,
to cover all gfx9 ASICs */
@ -4148,6 +4301,23 @@ static const struct soc15_reg_entry vgpr_init_regs[] = {
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
};
static const struct soc15_reg_entry vgpr_init_regs_arcturus[] = {
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 4 },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x81 },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x400000 }, /* 64KB LDS */
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff },
};
static const struct soc15_reg_entry sgpr1_init_regs[] = {
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 },
{ SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 },
@ -4278,7 +4448,10 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
adev->gfx.config.max_cu_per_sh *
adev->gfx.config.max_sh_per_se;
int sgpr_work_group_size = 5;
int gpr_reg_size = compute_dim_x / 16 + 6;
int gpr_reg_size = adev->gfx.config.max_shader_engines + 6;
int vgpr_init_shader_size;
const u32 *vgpr_init_shader_ptr;
const struct soc15_reg_entry *vgpr_init_regs_ptr;
/* only support when RAS is enabled */
if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
@ -4288,6 +4461,16 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
if (!ring->sched.ready)
return 0;
if (adev->asic_type == CHIP_ARCTURUS) {
vgpr_init_shader_ptr = vgpr_init_compute_shader_arcturus;
vgpr_init_shader_size = sizeof(vgpr_init_compute_shader_arcturus);
vgpr_init_regs_ptr = vgpr_init_regs_arcturus;
} else {
vgpr_init_shader_ptr = vgpr_init_compute_shader;
vgpr_init_shader_size = sizeof(vgpr_init_compute_shader);
vgpr_init_regs_ptr = vgpr_init_regs;
}
total_size =
(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* VGPRS */
total_size +=
@ -4296,7 +4479,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
(gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS2 */
total_size = ALIGN(total_size, 256);
vgpr_offset = total_size;
total_size += ALIGN(sizeof(vgpr_init_compute_shader), 256);
total_size += ALIGN(vgpr_init_shader_size, 256);
sgpr_offset = total_size;
total_size += sizeof(sgpr_init_compute_shader);
@ -4309,8 +4492,8 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
}
/* load the compute shaders */
for (i = 0; i < ARRAY_SIZE(vgpr_init_compute_shader); i++)
ib.ptr[i + (vgpr_offset / 4)] = vgpr_init_compute_shader[i];
for (i = 0; i < vgpr_init_shader_size/sizeof(u32); i++)
ib.ptr[i + (vgpr_offset / 4)] = vgpr_init_shader_ptr[i];
for (i = 0; i < ARRAY_SIZE(sgpr_init_compute_shader); i++)
ib.ptr[i + (sgpr_offset / 4)] = sgpr_init_compute_shader[i];
@ -4322,9 +4505,9 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
/* write the register state for the compute dispatch */
for (i = 0; i < gpr_reg_size; i++) {
ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1);
ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(vgpr_init_regs[i])
ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(vgpr_init_regs_ptr[i])
- PACKET3_SET_SH_REG_START;
ib.ptr[ib.length_dw++] = vgpr_init_regs[i].reg_value;
ib.ptr[ib.length_dw++] = vgpr_init_regs_ptr[i].reg_value;
}
/* write the shader start address: mmCOMPUTE_PGM_LO, mmCOMPUTE_PGM_HI */
gpu_addr = (ib.gpu_addr + (u64)vgpr_offset) >> 8;
@ -4336,7 +4519,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev)
/* write dispatch packet */
ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3);
ib.ptr[ib.length_dw++] = compute_dim_x; /* x */
ib.ptr[ib.length_dw++] = compute_dim_x * 2; /* x */
ib.ptr[ib.length_dw++] = 1; /* y */
ib.ptr[ib.length_dw++] = 1; /* z */
ib.ptr[ib.length_dw++] =
@ -4783,6 +4966,35 @@ static void gfx_v9_0_update_spm_vmid(struct amdgpu_device *adev, unsigned vmid)
WREG32_SOC15(GC, 0, mmRLC_SPM_MC_CNTL, data);
}
static bool gfx_v9_0_check_rlcg_range(struct amdgpu_device *adev,
uint32_t offset,
struct soc15_reg_rlcg *entries, int arr_size)
{
int i;
uint32_t reg;
if (!entries)
return false;
for (i = 0; i < arr_size; i++) {
const struct soc15_reg_rlcg *entry;
entry = &entries[i];
reg = adev->reg_offset[entry->hwip][entry->instance][entry->segment] + entry->reg;
if (offset == reg)
return true;
}
return false;
}
static bool gfx_v9_0_is_rlcg_access_range(struct amdgpu_device *adev, u32 offset)
{
return gfx_v9_0_check_rlcg_range(adev, offset,
(void *)rlcg_access_gc_9_0,
ARRAY_SIZE(rlcg_access_gc_9_0));
}
static const struct amdgpu_rlc_funcs gfx_v9_0_rlc_funcs = {
.is_rlc_enabled = gfx_v9_0_is_rlc_enabled,
.set_safe_mode = gfx_v9_0_set_safe_mode,
@ -4795,7 +5007,9 @@ static const struct amdgpu_rlc_funcs gfx_v9_0_rlc_funcs = {
.stop = gfx_v9_0_rlc_stop,
.reset = gfx_v9_0_rlc_reset,
.start = gfx_v9_0_rlc_start,
.update_spm_vmid = gfx_v9_0_update_spm_vmid
.update_spm_vmid = gfx_v9_0_update_spm_vmid,
.rlcg_wreg = gfx_v9_0_rlcg_wreg,
.is_rlcg_access_range = gfx_v9_0_is_rlcg_access_range,
};
static int gfx_v9_0_set_powergating_state(void *handle,
@ -6306,6 +6520,9 @@ static void gfx_v9_0_reset_ras_error_count(struct amdgpu_device *adev)
{
int i, j, k;
if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
return;
/* read back registers to clear the counters */
mutex_lock(&adev->grbm_idx_mutex);
for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {

3
drivers/gpu/drm/amd/amdgpu/gfx_v9_4.c
View File

@ -897,6 +897,9 @@ void gfx_v9_4_reset_ras_error_count(struct amdgpu_device *adev)
{
int i, j, k;
if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX))
return;
mutex_lock(&adev->grbm_idx_mutex);
for (i = 0; i < ARRAY_SIZE(gfx_v9_4_edc_counter_regs); i++) {
for (j = 0; j < gfx_v9_4_edc_counter_regs[i].se_num; j++) {

36
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
View File

@ -922,30 +922,20 @@ static int gmc_v9_0_late_init(void *handle)
if (r)
return r;
/* Check if ecc is available */
if (!amdgpu_sriov_vf(adev)) {
switch (adev->asic_type) {
case CHIP_VEGA10:
case CHIP_VEGA20:
case CHIP_ARCTURUS:
r = amdgpu_atomfirmware_mem_ecc_supported(adev);
if (!r) {
DRM_INFO("ECC is not present.\n");
if (adev->df.funcs->enable_ecc_force_par_wr_rmw)
adev->df.funcs->enable_ecc_force_par_wr_rmw(adev, false);
} else {
DRM_INFO("ECC is active.\n");
}
if (!amdgpu_sriov_vf(adev) && (adev->asic_type == CHIP_VEGA10)) {
r = amdgpu_atomfirmware_mem_ecc_supported(adev);
if (!r) {
DRM_INFO("ECC is not present.\n");
if (adev->df.funcs->enable_ecc_force_par_wr_rmw)
adev->df.funcs->enable_ecc_force_par_wr_rmw(adev, false);
} else
DRM_INFO("ECC is active.\n");
r = amdgpu_atomfirmware_sram_ecc_supported(adev);
if (!r) {
DRM_INFO("SRAM ECC is not present.\n");
} else {
DRM_INFO("SRAM ECC is active.\n");
}
break;
default:
break;
}
r = amdgpu_atomfirmware_sram_ecc_supported(adev);
if (!r)
DRM_INFO("SRAM ECC is not present.\n");
else
DRM_INFO("SRAM ECC is active.\n");
}
if (adev->mmhub.funcs && adev->mmhub.funcs->reset_ras_error_count)

2
drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c
View File

@ -693,7 +693,7 @@ static int jpeg_v2_0_set_clockgating_state(void *handle,
bool enable = (state == AMD_CG_STATE_GATE);
if (enable) {
if (jpeg_v2_0_is_idle(handle))
if (!jpeg_v2_0_is_idle(handle))
return -EBUSY;
jpeg_v2_0_enable_clock_gating(adev);
} else {

2
drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c
View File

@ -477,7 +477,7 @@ static int jpeg_v2_5_set_clockgating_state(void *handle,
continue;
if (enable) {
if (jpeg_v2_5_is_idle(handle))
if (!jpeg_v2_5_is_idle(handle))
return -EBUSY;
jpeg_v2_5_enable_clock_gating(adev, i);
} else {

1
drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
View File

@ -246,6 +246,7 @@ enum psp_gfx_fw_type {
GFX_FW_TYPE_SDMA6 = 56, /* SDMA6 MI */
GFX_FW_TYPE_SDMA7 = 57, /* SDMA7 MI */
GFX_FW_TYPE_VCN1 = 58, /* VCN1 MI */
GFX_FW_TYPE_CAP = 62, /* CAP_FW VG */
GFX_FW_TYPE_MAX
};

6
drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
View File

@ -26,6 +26,7 @@
#include "amdgpu.h"
#include "amdgpu_psp.h"
#include "amdgpu_ras.h"
#include "amdgpu_ucode.h"
#include "soc15_common.h"
#include "psp_v11_0.h"
@ -868,6 +869,11 @@ static int psp_v11_0_ras_trigger_error(struct psp_context *psp,
if (ret)
return -EINVAL;
/* If err_event_athub occurs error inject was successful, however
return status from TA is no long reliable */
if (amdgpu_ras_intr_triggered())
return 0;
return ras_cmd->ras_status;
}

24
drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
View File

@ -44,6 +44,7 @@
MODULE_FIRMWARE("amdgpu/vega10_sos.bin");
MODULE_FIRMWARE("amdgpu/vega10_asd.bin");
MODULE_FIRMWARE("amdgpu/vega10_cap.bin");
MODULE_FIRMWARE("amdgpu/vega12_sos.bin");
MODULE_FIRMWARE("amdgpu/vega12_asd.bin");
@ -63,6 +64,7 @@ static int psp_v3_1_init_microcode(struct psp_context *psp)
char fw_name[30];
int err = 0;
const struct psp_firmware_header_v1_0 *hdr;
struct amdgpu_firmware_info *info = NULL;
DRM_DEBUG("\n");
@ -112,6 +114,26 @@ static int psp_v3_1_init_microcode(struct psp_context *psp)
adev->psp.asd_start_addr = (uint8_t *)hdr +
le32_to_cpu(hdr->header.ucode_array_offset_bytes);
if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_VEGA10) {
snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_cap.bin",
chip_name);
err = request_firmware(&adev->psp.cap_fw, fw_name, adev->dev);
if (err)
goto out;
err = amdgpu_ucode_validate(adev->psp.cap_fw);
if (err)
goto out;
info = &adev->firmware.ucode[AMDGPU_UCODE_ID_CAP];
info->ucode_id = AMDGPU_UCODE_ID_CAP;
info->fw = adev->psp.cap_fw;
hdr = (const struct psp_firmware_header_v1_0 *)
adev->psp.cap_fw->data;
adev->firmware.fw_size += ALIGN(
le32_to_cpu(hdr->header.ucode_size_bytes), PAGE_SIZE);
}
return 0;
out:
if (err) {
@ -122,6 +144,8 @@ out:
adev->psp.sos_fw = NULL;
release_firmware(adev->psp.asd_fw);
adev->psp.asd_fw = NULL;
release_firmware(adev->psp.cap_fw);
adev->psp.cap_fw = NULL;
}
return err;

14
drivers/gpu/drm/amd/amdgpu/smu_v11_0_i2c.c
View File

@ -46,8 +46,7 @@
#define I2C_NO_STOP 1
#define I2C_RESTART 2
#define to_amdgpu_device(x) (container_of(x, struct amdgpu_ras, eeprom_control.eeprom_accessor))->adev
#define to_eeprom_control(x) container_of(x, struct amdgpu_ras_eeprom_control, eeprom_accessor)
#define to_amdgpu_device(x) (container_of(x, struct amdgpu_device, pm.smu_i2c))
static void smu_v11_0_i2c_set_clock_gating(struct i2c_adapter *control, bool en)
{
@ -592,7 +591,8 @@ static uint32_t smu_v11_0_i2c_eeprom_write_data(struct i2c_adapter *control,
static void lock_bus(struct i2c_adapter *i2c, unsigned int flags)
{
struct amdgpu_ras_eeprom_control *control = to_eeprom_control(i2c);
struct amdgpu_device *adev = to_amdgpu_device(i2c);
struct amdgpu_ras_eeprom_control *control = &adev->psp.ras.ras->eeprom_control;
if (!smu_v11_0_i2c_bus_lock(i2c)) {
DRM_ERROR("Failed to lock the bus from SMU");
@ -610,7 +610,8 @@ static int trylock_bus(struct i2c_adapter *i2c, unsigned int flags)
static void unlock_bus(struct i2c_adapter *i2c, unsigned int flags)
{
struct amdgpu_ras_eeprom_control *control = to_eeprom_control(i2c);
struct amdgpu_device *adev = to_amdgpu_device(i2c);
struct amdgpu_ras_eeprom_control *control = &adev->psp.ras.ras->eeprom_control;
if (!smu_v11_0_i2c_bus_unlock(i2c)) {
DRM_ERROR("Failed to unlock the bus from SMU");
@ -630,7 +631,8 @@ static int smu_v11_0_i2c_eeprom_i2c_xfer(struct i2c_adapter *i2c_adap,
struct i2c_msg *msgs, int num)
{
int i, ret;
struct amdgpu_ras_eeprom_control *control = to_eeprom_control(i2c_adap);
struct amdgpu_device *adev = to_amdgpu_device(i2c_adap);
struct amdgpu_ras_eeprom_control *control = &adev->psp.ras.ras->eeprom_control;
if (!control->bus_locked) {
DRM_ERROR("I2C bus unlocked, stopping transaction!");
@ -679,7 +681,7 @@ int smu_v11_0_i2c_eeprom_control_init(struct i2c_adapter *control)
control->class = I2C_CLASS_SPD;
control->dev.parent = &adev->pdev->dev;
control->algo = &smu_v11_0_i2c_eeprom_i2c_algo;
snprintf(control->name, sizeof(control->name), "RAS EEPROM");
snprintf(control->name, sizeof(control->name), "AMDGPU EEPROM");
control->lock_ops = &smu_v11_0_i2c_i2c_lock_ops;
res = i2c_add_adapter(control);

7
drivers/gpu/drm/amd/amdgpu/soc15.h
View File

@ -42,6 +42,13 @@ struct soc15_reg_golden {
u32 or_mask;
};
struct soc15_reg_rlcg {
u32 hwip;
u32 instance;
u32 segment;
u32 reg;
};
struct soc15_reg_entry {
uint32_t hwip;
uint32_t inst;

5
drivers/gpu/drm/amd/amdgpu/soc15_common.h
View File

@ -70,10 +70,9 @@
} \
} while (0)
#define AMDGPU_VIRT_SUPPORT_RLC_PRG_REG(a) (amdgpu_sriov_vf((a)) && !amdgpu_sriov_runtime((a)))
#define WREG32_RLC(reg, value) \
do { \
if (AMDGPU_VIRT_SUPPORT_RLC_PRG_REG(adev)) { \
if (amdgpu_sriov_fullaccess(adev)) { \
uint32_t i = 0; \
uint32_t retries = 50000; \
uint32_t r0 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG0_BASE_IDX] + mmSCRATCH_REG0; \
@ -98,7 +97,7 @@
#define WREG32_SOC15_RLC_SHADOW(ip, inst, reg, value) \
do { \
uint32_t target_reg = adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg;\
if (AMDGPU_VIRT_SUPPORT_RLC_PRG_REG(adev)) { \
if (amdgpu_sriov_fullaccess(adev)) { \
uint32_t r2 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG2; \
uint32_t r3 = adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3; \
uint32_t grbm_cntl = adev->reg_offset[GC_HWIP][0][mmGRBM_GFX_CNTL_BASE_IDX] + mmGRBM_GFX_CNTL; \

2
drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
View File

@ -1352,7 +1352,7 @@ static int vcn_v1_0_set_clockgating_state(void *handle,
if (enable) {
/* wait for STATUS to clear */
if (vcn_v1_0_is_idle(handle))
if (!vcn_v1_0_is_idle(handle))
return -EBUSY;
vcn_v1_0_enable_clock_gating(adev);
} else {

2
drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c
View File

@ -1251,7 +1251,7 @@ static int vcn_v2_0_set_clockgating_state(void *handle,
if (enable) {
/* wait for STATUS to clear */
if (vcn_v2_0_is_idle(handle))
if (!vcn_v2_0_is_idle(handle))
return -EBUSY;
vcn_v2_0_enable_clock_gating(adev);
} else {

37
drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c
View File

@ -74,29 +74,30 @@ static int amdgpu_ih_clientid_vcns[] = {
static int vcn_v2_5_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (adev->asic_type == CHIP_ARCTURUS) {
u32 harvest;
int i;
adev->vcn.num_vcn_inst = VCN25_MAX_HW_INSTANCES_ARCTURUS;
for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
harvest = RREG32_SOC15(UVD, i, mmCC_UVD_HARVESTING);
if (harvest & CC_UVD_HARVESTING__UVD_DISABLE_MASK)
adev->vcn.harvest_config |= 1 << i;
}
if (adev->vcn.harvest_config == (AMDGPU_VCN_HARVEST_VCN0 |
AMDGPU_VCN_HARVEST_VCN1))
/* both instances are harvested, disable the block */
return -ENOENT;
} else
adev->vcn.num_vcn_inst = 1;
if (amdgpu_sriov_vf(adev)) {
adev->vcn.num_vcn_inst = 2;
adev->vcn.harvest_config = 0;
adev->vcn.num_enc_rings = 1;
} else {
if (adev->asic_type == CHIP_ARCTURUS) {
u32 harvest;
int i;
adev->vcn.num_vcn_inst = VCN25_MAX_HW_INSTANCES_ARCTURUS;
for (i = 0; i < adev->vcn.num_vcn_inst; i++) {
harvest = RREG32_SOC15(UVD, i, mmCC_UVD_HARVESTING);
if (harvest & CC_UVD_HARVESTING__UVD_DISABLE_MASK)
adev->vcn.harvest_config |= 1 << i;
}
if (adev->vcn.harvest_config == (AMDGPU_VCN_HARVEST_VCN0 |
AMDGPU_VCN_HARVEST_VCN1))
/* both instances are harvested, disable the block */
return -ENOENT;
} else
adev->vcn.num_vcn_inst = 1;
adev->vcn.num_enc_rings = 2;
}
@ -1672,7 +1673,7 @@ static int vcn_v2_5_set_clockgating_state(void *handle,
return 0;
if (enable) {
if (vcn_v2_5_is_idle(handle))
if (!vcn_v2_5_is_idle(handle))
return -EBUSY;
vcn_v2_5_enable_clock_gating(adev);
} else {

2
drivers/gpu/drm/amd/amdkfd/kfd_flat_memory.c
View File

@ -316,7 +316,7 @@ static void kfd_init_apertures_vi(struct kfd_process_device *pdd, uint8_t id)
{
/*
* node id couldn't be 0 - the three MSB bits of
* aperture shoudn't be 0
* aperture shouldn't be 0
*/
pdd->lds_base = MAKE_LDS_APP_BASE_VI();
pdd->lds_limit = MAKE_LDS_APP_LIMIT(pdd->lds_base);

58
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
View File

@ -132,9 +132,6 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev);
/* removes and deallocates the drm structures, created by the above function */
static void amdgpu_dm_destroy_drm_device(struct amdgpu_display_manager *dm);
static void
amdgpu_dm_update_connector_after_detect(struct amdgpu_dm_connector *aconnector);
static int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm,
struct drm_plane *plane,
unsigned long possible_crtcs,
@ -410,8 +407,9 @@ static void dm_vupdate_high_irq(void *interrupt_params)
if (acrtc) {
acrtc_state = to_dm_crtc_state(acrtc->base.state);
DRM_DEBUG_DRIVER("crtc:%d, vupdate-vrr:%d\n", acrtc->crtc_id,
amdgpu_dm_vrr_active(acrtc_state));
DRM_DEBUG_VBL("crtc:%d, vupdate-vrr:%d\n",
acrtc->crtc_id,
amdgpu_dm_vrr_active(acrtc_state));
/* Core vblank handling is done here after end of front-porch in
* vrr mode, as vblank timestamping will give valid results
@ -461,8 +459,9 @@ static void dm_crtc_high_irq(void *interrupt_params)
if (acrtc) {
acrtc_state = to_dm_crtc_state(acrtc->base.state);
DRM_DEBUG_DRIVER("crtc:%d, vupdate-vrr:%d\n", acrtc->crtc_id,
amdgpu_dm_vrr_active(acrtc_state));
DRM_DEBUG_VBL("crtc:%d, vupdate-vrr:%d\n",
acrtc->crtc_id,
amdgpu_dm_vrr_active(acrtc_state));
/* Core vblank handling at start of front-porch is only possible
* in non-vrr mode, as only there vblank timestamping will give
@ -525,8 +524,8 @@ static void dm_dcn_crtc_high_irq(void *interrupt_params)
acrtc_state = to_dm_crtc_state(acrtc->base.state);
DRM_DEBUG_DRIVER("crtc:%d, vupdate-vrr:%d\n", acrtc->crtc_id,
amdgpu_dm_vrr_active(acrtc_state));
DRM_DEBUG_VBL("crtc:%d, vupdate-vrr:%d\n", acrtc->crtc_id,
amdgpu_dm_vrr_active(acrtc_state));
amdgpu_dm_crtc_handle_crc_irq(&acrtc->base);
drm_crtc_handle_vblank(&acrtc->base);
@ -1895,8 +1894,8 @@ static void update_connector_ext_caps(struct amdgpu_dm_connector *aconnector)
caps->aux_min_input_signal = min;
}
static void
amdgpu_dm_update_connector_after_detect(struct amdgpu_dm_connector *aconnector)
void amdgpu_dm_update_connector_after_detect(
struct amdgpu_dm_connector *aconnector)
{
struct drm_connector *connector = &aconnector->base;
struct drm_device *dev = connector->dev;
@ -2225,10 +2224,10 @@ static void handle_hpd_rx_irq(void *param)
}
}
#ifdef CONFIG_DRM_AMD_DC_HDCP
if (hpd_irq_data.bytes.device_service_irq.bits.CP_IRQ) {
if (adev->dm.hdcp_workqueue)
hdcp_handle_cpirq(adev->dm.hdcp_workqueue, aconnector->base.index);
}
if (hpd_irq_data.bytes.device_service_irq.bits.CP_IRQ) {
if (adev->dm.hdcp_workqueue)
hdcp_handle_cpirq(adev->dm.hdcp_workqueue, aconnector->base.index);
}
#endif
if ((dc_link->cur_link_settings.lane_count != LANE_COUNT_UNKNOWN) ||
(dc_link->type == dc_connection_mst_branch))
@ -3023,6 +3022,9 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev)
if (adev->asic_type != CHIP_CARRIZO && adev->asic_type != CHIP_STONEY)
dm->dc->debug.disable_stutter = amdgpu_pp_feature_mask & PP_STUTTER_MODE ? false : true;
/* No userspace support. */
dm->dc->debug.disable_tri_buf = true;
return 0;
fail:
kfree(aencoder);
@ -4314,9 +4316,22 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector,
struct dmcu *dmcu = core_dc->res_pool->dmcu;
stream->psr_version = dmcu->dmcu_version.psr_version;
mod_build_vsc_infopacket(stream,
&stream->vsc_infopacket,
&stream->use_vsc_sdp_for_colorimetry);
//
// should decide stream support vsc sdp colorimetry capability
// before building vsc info packet
//
stream->use_vsc_sdp_for_colorimetry = false;
if (aconnector->dc_sink->sink_signal == SIGNAL_TYPE_DISPLAY_PORT_MST) {
stream->use_vsc_sdp_for_colorimetry =
aconnector->dc_sink->is_vsc_sdp_colorimetry_supported;
} else {
if (stream->link->dpcd_caps.dpcd_rev.raw >= 0x14 &&
stream->link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED) {
stream->use_vsc_sdp_for_colorimetry = true;
}
}
mod_build_vsc_infopacket(stream, &stream->vsc_infopacket);
}
}
finish:
@ -6538,7 +6553,6 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
uint32_t target_vblank, last_flip_vblank;
bool vrr_active = amdgpu_dm_vrr_active(acrtc_state);
bool pflip_present = false;
bool swizzle = true;
struct {
struct dc_surface_update surface_updates[MAX_SURFACES];
struct dc_plane_info plane_infos[MAX_SURFACES];
@ -6584,9 +6598,6 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
dc_plane = dm_new_plane_state->dc_state;
if (dc_plane && !dc_plane->tiling_info.gfx9.swizzle)
swizzle = false;
bundle->surface_updates[planes_count].surface = dc_plane;
if (new_pcrtc_state->color_mgmt_changed) {
bundle->surface_updates[planes_count].gamma = dc_plane->gamma_correction;
@ -6795,8 +6806,7 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
amdgpu_dm_link_setup_psr(acrtc_state->stream);
else if ((acrtc_state->update_type == UPDATE_TYPE_FAST) &&
acrtc_state->stream->link->psr_feature_enabled &&
!acrtc_state->stream->link->psr_allow_active &&
swizzle) {
!acrtc_state->stream->link->psr_allow_active) {
amdgpu_dm_psr_enable(acrtc_state->stream);
}

3
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
View File

@ -483,6 +483,9 @@ int amdgpu_dm_update_crtc_color_mgmt(struct dm_crtc_state *crtc);
int amdgpu_dm_update_plane_color_mgmt(struct dm_crtc_state *crtc,
struct dc_plane_state *dc_plane_state);
void amdgpu_dm_update_connector_after_detect(
struct amdgpu_dm_connector *aconnector);
extern const struct drm_encoder_helper_funcs amdgpu_dm_encoder_helper_funcs;
#endif /* __AMDGPU_DM_H__ */

14
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
View File

@ -580,6 +580,20 @@ enum dc_edid_status dm_helpers_read_local_edid(
/* We don't need the original edid anymore */
kfree(edid);
/* connector->display_info will be parsed from EDID and saved
* into drm_connector->display_info from edid by call stack
* below:
* drm_parse_ycbcr420_deep_color_info
* drm_parse_hdmi_forum_vsdb
* drm_parse_cea_ext
* drm_add_display_info
* drm_connector_update_edid_property
*
* drm_connector->display_info will be used by amdgpu_dm funcs,
* like fill_stream_properties_from_drm_display_mode
*/
amdgpu_dm_update_connector_after_detect(aconnector);
edid_status = dm_helpers_parse_edid_caps(
ctx,
&sink->dc_edid,

4
drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
View File

@ -267,7 +267,7 @@ static struct atom_display_object_path_v2 *get_bios_object(
&& id.enum_id == obj_id.enum_id)
return &bp->object_info_tbl.v1_4->display_path[i];
}
/* fall through */
fallthrough;
case OBJECT_TYPE_CONNECTOR:
case OBJECT_TYPE_GENERIC:
/* Both Generic and Connector Object ID
@ -280,7 +280,7 @@ static struct atom_display_object_path_v2 *get_bios_object(
&& id.enum_id == obj_id.enum_id)
return &bp->object_info_tbl.v1_4->display_path[i];
}
/* fall through */
fallthrough;
default:
return NULL;
}

5
drivers/gpu/drm/amd/display/dc/clk_mgr/dcn20/dcn20_clk_mgr.c
View File

@ -115,12 +115,11 @@ void dcn20_update_clocks_update_dpp_dto(struct clk_mgr_internal *clk_mgr,
dpp_inst = i;
dppclk_khz = context->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
prev_dppclk_khz = clk_mgr->base.ctx->dc->current_state->res_ctx.pipe_ctx[i].plane_res.bw.dppclk_khz;
prev_dppclk_khz = clk_mgr->dccg->pipe_dppclk_khz[i];
if ((prev_dppclk_khz > dppclk_khz && safe_to_lower) || prev_dppclk_khz < dppclk_khz) {
if (safe_to_lower || prev_dppclk_khz < dppclk_khz)
clk_mgr->dccg->funcs->update_dpp_dto(
clk_mgr->dccg, dpp_inst, dppclk_khz);
}
}
}

15
drivers/gpu/drm/amd/display/dc/core/dc.c
View File

@ -1365,7 +1365,7 @@ bool dc_post_update_surfaces_to_stream(struct dc *dc)
int i;
struct dc_state *context = dc->current_state;
if ((!dc->clk_optimized_required && !dc->wm_optimized_required) || dc->optimize_seamless_boot_streams > 0)
if ((!dc->optimized_required) || dc->optimize_seamless_boot_streams > 0)
return true;
post_surface_trace(dc);
@ -1379,7 +1379,7 @@ bool dc_post_update_surfaces_to_stream(struct dc *dc)
dc->hwss.optimize_bandwidth(dc, context);
dc->clk_optimized_required = false;
dc->optimized_required = false;
dc->wm_optimized_required = false;
return true;
@ -1828,11 +1828,12 @@ enum surface_update_type dc_check_update_surfaces_for_stream(
// If there's an available clock comparator, we use that.
if (dc->clk_mgr->funcs->are_clock_states_equal) {
if (!dc->clk_mgr->funcs->are_clock_states_equal(&dc->clk_mgr->clks, &dc->current_state->bw_ctx.bw.dcn.clk))
dc->clk_optimized_required = true;
dc->optimized_required = true;
// Else we fallback to mem compare.
} else if (memcmp(&dc->current_state->bw_ctx.bw.dcn.clk, &dc->clk_mgr->clks, offsetof(struct dc_clocks, prev_p_state_change_support)) != 0) {
dc->clk_optimized_required = true;
}
dc->optimized_required = true;
} else if (dc->wm_optimized_required)
dc->optimized_required = true;
}
return type;
@ -1871,6 +1872,8 @@ static void copy_surface_update_to_plane(
surface->time.index++;
if (surface->time.index >= DC_PLANE_UPDATE_TIMES_MAX)
surface->time.index = 0;
surface->triplebuffer_flips = srf_update->flip_addr->triplebuffer_flips;
}
if (srf_update->scaling_info) {
@ -2202,7 +2205,7 @@ static void commit_planes_for_stream(struct dc *dc,
dc->optimize_seamless_boot_streams--;
if (dc->optimize_seamless_boot_streams == 0)
dc->clk_optimized_required = true;
dc->optimized_required = true;
}
}

11
drivers/gpu/drm/amd/display/dc/core/dc_link.c
View File

@ -3069,14 +3069,9 @@ void core_link_enable_stream(
if (pipe_ctx->stream->timing.flags.DSC) {
if (dc_is_dp_signal(pipe_ctx->stream->signal) ||
dc_is_virtual_signal(pipe_ctx->stream->signal)) {
/* Here we only need to enable DSC on RX. DSC HW programming
* was done earlier, as part of timing programming.
*/
dp_set_dsc_on_rx(pipe_ctx, true);
}
dc_is_virtual_signal(pipe_ctx->stream->signal))
dp_set_dsc_enable(pipe_ctx, true);
}
dc->hwss.enable_stream(pipe_ctx);
/* Set DPS PPS SDP (AKA "info frames") */
@ -3103,7 +3098,7 @@ void core_link_enable_stream(
} else { // if (IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment))
if (dc_is_dp_signal(pipe_ctx->stream->signal) ||
dc_is_virtual_signal(pipe_ctx->stream->signal))
dp_set_dsc_on_rx(pipe_ctx, true);
dp_set_dsc_enable(pipe_ctx, true);
}
}

20
drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c
View File

@ -2674,9 +2674,12 @@ static void dp_test_send_link_test_pattern(struct dc_link *link)
break;
}
test_pattern_color_space = dpcd_test_params.bits.YCBCR_COEFS ?
DP_TEST_PATTERN_COLOR_SPACE_YCBCR709 :
DP_TEST_PATTERN_COLOR_SPACE_YCBCR601;
if (dpcd_test_params.bits.CLR_FORMAT == 0)
test_pattern_color_space = DP_TEST_PATTERN_COLOR_SPACE_RGB;
else
test_pattern_color_space = dpcd_test_params.bits.YCBCR_COEFS ?
DP_TEST_PATTERN_COLOR_SPACE_YCBCR709 :
DP_TEST_PATTERN_COLOR_SPACE_YCBCR601;
dc_link_dp_set_test_pattern(
link,
@ -3438,6 +3441,17 @@ static bool retrieve_link_cap(struct dc_link *link)
sink_id.ieee_device_id,
sizeof(sink_id.ieee_device_id));
/* Quirk Apple MBP 2017 15" Retina panel: Wrong DP_MAX_LINK_RATE */
{
uint8_t str_mbp_2017[] = { 101, 68, 21, 101, 98, 97 };
if ((link->dpcd_caps.sink_dev_id == 0x0010fa) &&
!memcmp(link->dpcd_caps.sink_dev_id_str, str_mbp_2017,
sizeof(str_mbp_2017))) {
link->reported_link_cap.link_rate = 0x0c;
}
}
core_link_read_dpcd(
link,
DP_SINK_HW_REVISION_START,

2
drivers/gpu/drm/amd/display/dc/core/dc_link_hwss.c
View File

@ -394,7 +394,7 @@ static void dsc_optc_config_log(struct display_stream_compressor *dsc,
DC_LOG_DSC("\tslice_width %d", config->slice_width);
}
bool dp_set_dsc_on_rx(struct pipe_ctx *pipe_ctx, bool enable)
static bool dp_set_dsc_on_rx(struct pipe_ctx *pipe_ctx, bool enable)
{
struct dc *dc = pipe_ctx->stream->ctx->dc;
struct dc_stream_state *stream = pipe_ctx->stream;

7
drivers/gpu/drm/amd/display/dc/core/dc_resource.c
View File

@ -893,6 +893,7 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx)
int vpc_div = (data->format == PIXEL_FORMAT_420BPP8
|| data->format == PIXEL_FORMAT_420BPP10) ? 2 : 1;
bool orthogonal_rotation, flip_vert_scan_dir, flip_horz_scan_dir;
int odm_idx = 0;
/*
* Need to calculate the scan direction for viewport to make adjustments
@ -924,11 +925,13 @@ static void calculate_inits_and_adj_vp(struct pipe_ctx *pipe_ctx)
* stream->dst.width / stream->src.width -
src.x * plane_state->dst_rect.width / src.width
* stream->dst.width / stream->src.width);
/*modified recout_skip_h calculation due to odm having no recout offset caused by split*/
/*modified recout_skip_h calculation due to odm having no recout offset*/
while (odm_pipe) {
recout_skip_h += odm_pipe->plane_res.scl_data.recout.width + odm_pipe->plane_res.scl_data.recout.x;
odm_idx++;
odm_pipe = odm_pipe->prev_odm_pipe;
}
if (odm_idx)
recout_skip_h += odm_idx * data->recout.width;
recout_skip_v = data->recout.y - (stream->dst.y + (plane_state->dst_rect.y - stream->src.y)
* stream->dst.height / stream->src.height -

5
drivers/gpu/drm/amd/display/dc/dc.h
View File

@ -521,7 +521,7 @@ struct dc {
struct dce_hwseq *hwseq;
/* Require to optimize clocks and bandwidth for added/removed planes */
bool clk_optimized_required;
bool optimized_required;
bool wm_optimized_required;
/* Require to maintain clocks and bandwidth for UEFI enabled HW */
@ -872,6 +872,7 @@ struct dc_flip_addrs {
unsigned int flip_timestamp_in_us;
bool flip_immediate;
/* TODO: add flip duration for FreeSync */
bool triplebuffer_flips;
};
bool dc_post_update_surfaces_to_stream(
@ -1046,6 +1047,8 @@ struct dc_sink {
struct dc_sink_dsc_caps dsc_caps;
struct dc_sink_fec_caps fec_caps;
bool is_vsc_sdp_colorimetry_supported;
/* private to DC core */
struct dc_link *link;
struct dc_context *ctx;

2
drivers/gpu/drm/amd/display/dc/dce/dce_aux.c
View File

@ -645,7 +645,7 @@ bool dce_aux_transfer_with_retries(struct ddc_service *ddc,
case AUX_TRANSACTION_REPLY_AUX_DEFER:
case AUX_TRANSACTION_REPLY_I2C_OVER_AUX_DEFER:
retry_on_defer = true;
/* fall through */
fallthrough;
case AUX_TRANSACTION_REPLY_I2C_OVER_AUX_NACK:
if (++aux_defer_retries >= AUX_MAX_DEFER_RETRIES) {
goto fail;

23
drivers/gpu/drm/amd/display/dc/dce/dce_i2c_hw.c
View File

@ -267,6 +267,9 @@ static void set_speed(
uint32_t xtal_ref_div = 0;
uint32_t prescale = 0;
if (speed == 0)
return;
REG_GET(MICROSECOND_TIME_BASE_DIV, XTAL_REF_DIV, &xtal_ref_div);
if (xtal_ref_div == 0)
@ -274,17 +277,15 @@ static void set_speed(
prescale = ((dce_i2c_hw->reference_frequency * 2) / xtal_ref_div) / speed;
if (speed) {
if (dce_i2c_hw->masks->DC_I2C_DDC1_START_STOP_TIMING_CNTL)
REG_UPDATE_N(SPEED, 3,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE), prescale,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD), 2,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_START_STOP_TIMING_CNTL), speed > 50 ? 2:1);
else
REG_UPDATE_N(SPEED, 2,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE), prescale,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD), 2);
}
if (dce_i2c_hw->masks->DC_I2C_DDC1_START_STOP_TIMING_CNTL)
REG_UPDATE_N(SPEED, 3,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE), prescale,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD), 2,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_START_STOP_TIMING_CNTL), speed > 50 ? 2:1);
else
REG_UPDATE_N(SPEED, 2,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_PRESCALE), prescale,
FN(DC_I2C_DDC1_SPEED, DC_I2C_DDC1_THRESHOLD), 2);
}
static bool setup_engine(

2
drivers/gpu/drm/amd/display/dc/dce/dce_mem_input.c
View File

@ -479,7 +479,7 @@ static void program_grph_pixel_format(
case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F:
sign = 1;
floating = 1;
/* fall through */
fallthrough;
case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: /* shouldn't this get float too? */
case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616:
grph_depth = 3;

8
drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c
View File

@ -134,11 +134,9 @@ static bool dmub_psr_copy_settings(struct dmub_psr *dmub,
int i = 0;
for (i = 0; i < MAX_PIPES; i++) {
if (res_ctx &&
res_ctx->pipe_ctx[i].stream &&
res_ctx->pipe_ctx[i].stream->link &&
res_ctx->pipe_ctx[i].stream->link == link &&
res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) {
if (res_ctx->pipe_ctx[i].stream &&
res_ctx->pipe_ctx[i].stream->link == link &&
res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) {
pipe_ctx = &res_ctx->pipe_ctx[i];
break;
}

7
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
View File

@ -1048,7 +1048,7 @@ void dcn10_plane_atomic_disconnect(struct dc *dc, struct pipe_ctx *pipe_ctx)
if (opp != NULL)
opp->mpcc_disconnect_pending[pipe_ctx->plane_res.mpcc_inst] = true;
dc->clk_optimized_required = true;
dc->optimized_required = true;
if (hubp->funcs->hubp_disconnect)
hubp->funcs->hubp_disconnect(hubp);
@ -1099,7 +1099,7 @@ void dcn10_plane_atomic_disable(struct dc *dc, struct pipe_ctx *pipe_ctx)
false);
hubp->power_gated = true;
dc->clk_optimized_required = false; /* We're powering off, no need to optimize */
dc->optimized_required = false; /* We're powering off, no need to optimize */
hws->funcs.plane_atomic_power_down(dc,
pipe_ctx->plane_res.dpp,
@ -1356,6 +1356,9 @@ void dcn10_init_hw(struct dc *dc)
*/
if (dcb->funcs->is_accelerated_mode(dcb) || dc->config.power_down_display_on_boot) {
hws->funcs.init_pipes(dc, dc->current_state);
if (dc->res_pool->hubbub->funcs->allow_self_refresh_control)
dc->res_pool->hubbub->funcs->allow_self_refresh_control(dc->res_pool->hubbub,
!dc->res_pool->hubbub->ctx->dc->debug.disable_stutter);
}
for (i = 0; i < res_pool->audio_count; i++) {

2
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_link_encoder.h
View File

@ -62,11 +62,11 @@
SRI(DP_DPHY_FAST_TRAINING, DP, id), \
SRI(DP_SEC_CNTL1, DP, id), \
SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \
SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
SRI(DP_DPHY_HBR2_PATTERN_CONTROL, DP, id)
#define LE_DCN10_REG_LIST(id)\
SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \
LE_DCN_COMMON_REG_LIST(id)
struct dcn10_link_enc_aux_registers {

4
drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c
View File

@ -570,7 +570,7 @@ static const struct dc_plane_cap plane_cap = {
static const struct dc_debug_options debug_defaults_drv = {
.sanity_checks = true,
.disable_dmcu = true,
.disable_dmcu = false,
.force_abm_enable = false,
.timing_trace = false,
.clock_trace = true,
@ -598,7 +598,7 @@ static const struct dc_debug_options debug_defaults_drv = {
};
static const struct dc_debug_options debug_defaults_diags = {
.disable_dmcu = true,
.disable_dmcu = false,
.force_abm_enable = false,
.timing_trace = true,
.clock_trace = true,

2
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dccg.c
View File

@ -70,6 +70,8 @@ void dccg2_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk)
REG_UPDATE(DPPCLK_DTO_CTRL,
DPPCLK_DTO_ENABLE[dpp_inst], 0);
}
dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk;
}
void dccg2_get_dccg_ref_freq(struct dccg *dccg,

10
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
View File

@ -623,13 +623,6 @@ enum dc_status dcn20_enable_stream_timing(
/* TODO check if timing_changed, disable stream if timing changed */
/* Have to setup DSC here to make sure the bandwidth sent to DIG BE won't be bigger than
* what the link and/or DIG BE can handle. VBID[6]/CompressedStream_flag will be automatically
* set at a later time when the video is enabled (DP_VID_STREAM_EN = 1).
*/
if (pipe_ctx->stream->timing.flags.DSC)
dp_set_dsc_on_stream(pipe_ctx, true);
for (odm_pipe = pipe_ctx->next_odm_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe) {
opp_inst[opp_cnt] = odm_pipe->stream_res.opp->inst;
opp_cnt++;
@ -654,6 +647,9 @@ enum dc_status dcn20_enable_stream_timing(
return DC_ERROR_UNEXPECTED;
}
if (dc->hwseq->funcs.PLAT_58856_wa && (!dc_is_dp_signal(stream->signal)))
dc->hwseq->funcs.PLAT_58856_wa(context, pipe_ctx);
pipe_ctx->stream_res.tg->funcs->program_timing(
pipe_ctx->stream_res.tg,
&stream->timing,

1
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_init.c
View File

@ -111,7 +111,6 @@ static const struct hwseq_private_funcs dcn20_private_funcs = {
.enable_power_gating_plane = dcn20_enable_power_gating_plane,
.dpp_pg_control = dcn20_dpp_pg_control,
.hubp_pg_control = dcn20_hubp_pg_control,
.dsc_pg_control = NULL,
.update_odm = dcn20_update_odm,
.dsc_pg_control = dcn20_dsc_pg_control,
.get_surface_visual_confirm_color = dcn10_get_surface_visual_confirm_color,

99
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
View File

@ -1041,7 +1041,7 @@ static const struct resource_caps res_cap_nv14 = {
};
static const struct dc_debug_options debug_defaults_drv = {
.disable_dmcu = true,
.disable_dmcu = false,
.force_abm_enable = false,
.timing_trace = false,
.clock_trace = true,
@ -1060,7 +1060,7 @@ static const struct dc_debug_options debug_defaults_drv = {
};
static const struct dc_debug_options debug_defaults_diags = {
.disable_dmcu = true,
.disable_dmcu = false,
.force_abm_enable = false,
.timing_trace = true,
.clock_trace = true,
@ -1671,7 +1671,7 @@ static void acquire_dsc(struct resource_context *res_ctx,
}
}
static void release_dsc(struct resource_context *res_ctx,
void dcn20_release_dsc(struct resource_context *res_ctx,
const struct resource_pool *pool,
struct display_stream_compressor **dsc)
{
@ -1731,7 +1731,7 @@ static enum dc_status remove_dsc_from_stream_resource(struct dc *dc,
pipe_ctx = &new_ctx->res_ctx.pipe_ctx[i];
if (pipe_ctx->stream_res.dsc)
release_dsc(&new_ctx->res_ctx, dc->res_pool, &pipe_ctx->stream_res.dsc);
dcn20_release_dsc(&new_ctx->res_ctx, dc->res_pool, &pipe_ctx->stream_res.dsc);
}
}
@ -2502,7 +2502,7 @@ struct pipe_ctx *dcn20_find_secondary_pipe(struct dc *dc,
return secondary_pipe;
}
void dcn20_merge_pipes_for_validate(
static void dcn20_merge_pipes_for_validate(
struct dc *dc,
struct dc_state *context)
{
@ -2527,7 +2527,7 @@ void dcn20_merge_pipes_for_validate(
odm_pipe->prev_odm_pipe = NULL;
odm_pipe->next_odm_pipe = NULL;
if (odm_pipe->stream_res.dsc)
release_dsc(&context->res_ctx, dc->res_pool, &odm_pipe->stream_res.dsc);
dcn20_release_dsc(&context->res_ctx, dc->res_pool, &odm_pipe->stream_res.dsc);
/* Clear plane_res and stream_res */
memset(&odm_pipe->plane_res, 0, sizeof(odm_pipe->plane_res));
memset(&odm_pipe->stream_res, 0, sizeof(odm_pipe->stream_res));
@ -2565,41 +2565,29 @@ int dcn20_validate_apply_pipe_split_flags(
struct dc *dc,
struct dc_state *context,
int vlevel,
bool *split)
bool *split,
bool *merge)
{
int i, pipe_idx, vlevel_split;
int plane_count = 0;
bool force_split = false;
bool avoid_split = dc->debug.pipe_split_policy != MPC_SPLIT_DYNAMIC;
bool avoid_split = dc->debug.pipe_split_policy == MPC_SPLIT_AVOID;
/* Single display loop, exits if there is more than one display */
if (context->stream_count > 1) {
if (dc->debug.pipe_split_policy == MPC_SPLIT_AVOID_MULT_DISP)
avoid_split = true;
} else if (dc->debug.force_single_disp_pipe_split)
force_split = true;
/* TODO: fix dc bugs and remove this split threshold thing */
for (i = 0; i < dc->res_pool->pipe_count; i++) {
struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
bool exit_loop = false;
if (!pipe->stream || pipe->top_pipe)
continue;
if (dc->debug.force_single_disp_pipe_split) {
if (!force_split)
force_split = true;
else {
force_split = false;
exit_loop = true;
}
}
if (dc->debug.pipe_split_policy == MPC_SPLIT_AVOID_MULT_DISP) {
if (avoid_split)
avoid_split = false;
else {
avoid_split = true;
exit_loop = true;
}
}
if (exit_loop)
break;
if (pipe->stream && !pipe->prev_odm_pipe &&
(!pipe->top_pipe || pipe->top_pipe->plane_state != pipe->plane_state))
++plane_count;
}
/* TODO: fix dc bugs and remove this split threshold thing */
if (context->stream_count > dc->res_pool->pipe_count / 2)
if (plane_count > dc->res_pool->pipe_count / 2)
avoid_split = true;
/* Avoid split loop looks for lowest voltage level that allows most unsplit pipes possible */
@ -2622,11 +2610,12 @@ int dcn20_validate_apply_pipe_split_flags(
/* Split loop sets which pipe should be split based on dml outputs and dc flags */
for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
int pipe_plane = context->bw_ctx.dml.vba.pipe_plane[pipe_idx];
if (!context->res_ctx.pipe_ctx[i].stream)
continue;
if (force_split || context->bw_ctx.dml.vba.NoOfDPP[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx] > 1)
if (force_split || context->bw_ctx.dml.vba.NoOfDPP[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_plane] > 1)
split[i] = true;
if ((pipe->stream->view_format ==
VIEW_3D_FORMAT_SIDE_BY_SIDE ||
@ -2639,10 +2628,44 @@ int dcn20_validate_apply_pipe_split_flags(
split[i] = true;
if (dc->debug.force_odm_combine & (1 << pipe->stream_res.tg->inst)) {
split[i] = true;
context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx] = dm_odm_combine_mode_2to1;
context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_plane] = dm_odm_combine_mode_2to1;
}
context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_idx] =
context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_idx];
context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_plane] =
context->bw_ctx.dml.vba.ODMCombineEnablePerState[vlevel][pipe_plane];
if (pipe->prev_odm_pipe && context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_plane] != dm_odm_combine_mode_disabled) {
/*Already split odm pipe tree, don't try to split again*/
split[i] = false;
split[pipe->prev_odm_pipe->pipe_idx] = false;
} else if (pipe->top_pipe && pipe->plane_state == pipe->top_pipe->plane_state
&& context->bw_ctx.dml.vba.ODMCombineEnabled[pipe_plane] == dm_odm_combine_mode_disabled) {
/*Already split mpc tree, don't try to split again, assumes only 2x mpc combine*/
split[i] = false;
split[pipe->top_pipe->pipe_idx] = false;
} else if (pipe->prev_odm_pipe || (pipe->top_pipe && pipe->plane_state == pipe->top_pipe->plane_state)) {
if (split[i] == false) {
/*Exiting mpc/odm combine*/
merge[i] = true;
if (pipe->prev_odm_pipe) {
ASSERT(0); /*should not actually happen yet*/
merge[pipe->prev_odm_pipe->pipe_idx] = true;
} else
merge[pipe->top_pipe->pipe_idx] = true;
} else {
/*Transition from mpc combine to odm combine or vice versa*/
ASSERT(0); /*should not actually happen yet*/
split[i] = true;
merge[i] = true;
if (pipe->prev_odm_pipe) {
split[pipe->prev_odm_pipe->pipe_idx] = true;
merge[pipe->prev_odm_pipe->pipe_idx] = true;
} else {
split[pipe->top_pipe->pipe_idx] = true;
merge[pipe->top_pipe->pipe_idx] = true;
}
}
}
/* Adjust dppclk when split is forced, do not bother with dispclk */
if (split[i] && context->bw_ctx.dml.vba.NoOfDPP[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx] == 1)
context->bw_ctx.dml.vba.RequiredDPPCLK[vlevel][context->bw_ctx.dml.vba.maxMpcComb][pipe_idx] /= 2;
@ -2684,7 +2707,7 @@ bool dcn20_fast_validate_bw(
if (vlevel > context->bw_ctx.dml.soc.num_states)
goto validate_fail;
vlevel = dcn20_validate_apply_pipe_split_flags(dc, context, vlevel, split);
vlevel = dcn20_validate_apply_pipe_split_flags(dc, context, vlevel, split, NULL);
/*initialize pipe_just_split_from to invalid idx*/
for (i = 0; i < MAX_PIPES; i++)

9
drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.h
View File

@ -119,14 +119,15 @@ void dcn20_set_mcif_arb_params(
display_e2e_pipe_params_st *pipes,
int pipe_cnt);
bool dcn20_validate_bandwidth(struct dc *dc, struct dc_state *context, bool fast_validate);
void dcn20_merge_pipes_for_validate(
struct dc *dc,
struct dc_state *context);
int dcn20_validate_apply_pipe_split_flags(
struct dc *dc,
struct dc_state *context,
int vlevel,
bool *split);
bool *split,
bool *merge);
void dcn20_release_dsc(struct resource_context *res_ctx,
const struct resource_pool *pool,
struct display_stream_compressor **dsc);
bool dcn20_validate_dsc(struct dc *dc, struct dc_state *new_ctx);
void dcn20_split_stream_for_mpc(
struct resource_context *res_ctx,

22
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hwseq.c
View File

@ -112,3 +112,25 @@ void dcn21_optimize_pwr_state(
true);
}
/* If user hotplug a HDMI monitor while in monitor off,
* OS will do a mode set (with output timing) but keep output off.
* In this case DAL will ask vbios to power up the pll in the PHY.
* If user unplug the monitor (while we are on monitor off) or
* system attempt to enter modern standby (which we will disable PLL),
* PHY will hang on the next mode set attempt.
* if enable PLL follow by disable PLL (without executing lane enable/disable),
* RDPCS_PHY_DP_MPLLB_STATE remains 1,
* which indicate that PLL disable attempt actually didn’t go through.
* As a workaround, insert PHY lane enable/disable before PLL disable.
*/
void dcn21_PLAT_58856_wa(struct dc_state *context, struct pipe_ctx *pipe_ctx)
{
if (!pipe_ctx->stream->dpms_off)
return;
pipe_ctx->stream->dpms_off = false;
core_link_enable_stream(context, pipe_ctx);
core_link_disable_stream(pipe_ctx);
pipe_ctx->stream->dpms_off = true;
}

3
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_hwseq.h
View File

@ -44,4 +44,7 @@ void dcn21_optimize_pwr_state(
const struct dc *dc,
struct dc_state *context);
void dcn21_PLAT_58856_wa(struct dc_state *context,
struct pipe_ctx *pipe_ctx);
#endif /* __DC_HWSS_DCN21_H__ */

2
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_init.c
View File

@ -119,7 +119,6 @@ static const struct hwseq_private_funcs dcn21_private_funcs = {
.enable_power_gating_plane = dcn20_enable_power_gating_plane,
.dpp_pg_control = dcn20_dpp_pg_control,
.hubp_pg_control = dcn20_hubp_pg_control,
.dsc_pg_control = NULL,
.update_odm = dcn20_update_odm,
.dsc_pg_control = dcn20_dsc_pg_control,
.get_surface_visual_confirm_color = dcn10_get_surface_visual_confirm_color,
@ -131,6 +130,7 @@ static const struct hwseq_private_funcs dcn21_private_funcs = {
.dccg_init = dcn20_dccg_init,
.set_blend_lut = dcn20_set_blend_lut,
.set_shaper_3dlut = dcn20_set_shaper_3dlut,
.PLAT_58856_wa = dcn21_PLAT_58856_wa,
};
void dcn21_hw_sequencer_construct(struct dc *dc)

6
drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c
View File

@ -855,7 +855,7 @@ static const struct dc_plane_cap plane_cap = {
};
static const struct dc_debug_options debug_defaults_drv = {
.disable_dmcu = true,
.disable_dmcu = false,
.force_abm_enable = false,
.timing_trace = false,
.clock_trace = true,
@ -876,7 +876,7 @@ static const struct dc_debug_options debug_defaults_drv = {
};
static const struct dc_debug_options debug_defaults_diags = {
.disable_dmcu = true,
.disable_dmcu = false,
.force_abm_enable = false,
.timing_trace = true,
.clock_trace = true,
@ -1864,7 +1864,7 @@ static bool dcn21_resource_construct(
goto create_fail;
}
if (dc->config.psr_on_dmub) {
if (dc->debug.disable_dmcu) {
pool->base.psr = dmub_psr_create(ctx);
if (pool->base.psr == NULL) {

1
drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h
View File

@ -85,7 +85,6 @@ void dp_set_fec_enable(struct dc_link *link, bool enable);
bool dp_set_dsc_enable(struct pipe_ctx *pipe_ctx, bool enable);
bool dp_set_dsc_pps_sdp(struct pipe_ctx *pipe_ctx, bool enable);
void dp_set_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable);
bool dp_set_dsc_on_rx(struct pipe_ctx *pipe_ctx, bool enable);
bool dp_update_dsc_config(struct pipe_ctx *pipe_ctx);
#endif /* __DC_LINK_DP_H__ */

3
drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h
View File

@ -27,11 +27,12 @@
#define __DAL_DCCG_H__
#include "dc_types.h"
#include "hw_shared.h"
struct dccg {
struct dc_context *ctx;
const struct dccg_funcs *funcs;
int pipe_dppclk_khz[MAX_PIPES];
int ref_dppclk;
};

2
drivers/gpu/drm/amd/display/dc/inc/hw_sequencer_private.h
View File

@ -145,6 +145,8 @@ struct hwseq_private_funcs {
const struct dc_plane_state *plane_state);
bool (*set_shaper_3dlut)(struct pipe_ctx *pipe_ctx,
const struct dc_plane_state *plane_state);
void (*PLAT_58856_wa)(struct dc_state *context,
struct pipe_ctx *pipe_ctx);
};
struct dce_hwseq {

3
drivers/gpu/drm/amd/display/dmub/inc/dmub_srv.h
View File

@ -104,7 +104,7 @@ enum dmub_window_id {
DMUB_WINDOW_4_MAILBOX,
DMUB_WINDOW_5_TRACEBUFF,
DMUB_WINDOW_6_FW_STATE,
DMUB_WINDOW_7_RESERVED,
DMUB_WINDOW_7_SCRATCH_MEM,
DMUB_WINDOW_TOTAL,
};
@ -316,6 +316,7 @@ struct dmub_srv {
enum dmub_asic asic;
void *user_ctx;
bool is_virtual;
struct dmub_fb scratch_mem_fb;
volatile const struct dmub_fw_state *fw_state;
/* private: internal use only */

17
drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c
View File

@ -52,8 +52,11 @@
/* Default tracebuffer size if meta is absent. */
#define DMUB_TRACE_BUFFER_SIZE (1024)
/* Default scratch mem size. */
#define DMUB_SCRATCH_MEM_SIZE (256)
/* Number of windows in use. */
#define DMUB_NUM_WINDOWS (DMUB_WINDOW_6_FW_STATE + 1)
#define DMUB_NUM_WINDOWS (DMUB_WINDOW_TOTAL)
/* Base addresses. */
#define DMUB_CW0_BASE (0x60000000)
@ -211,9 +214,11 @@ dmub_srv_calc_region_info(struct dmub_srv *dmub,
struct dmub_region *mail = &out->regions[DMUB_WINDOW_4_MAILBOX];
struct dmub_region *trace_buff = &out->regions[DMUB_WINDOW_5_TRACEBUFF];
struct dmub_region *fw_state = &out->regions[DMUB_WINDOW_6_FW_STATE];
struct dmub_region *scratch_mem = &out->regions[DMUB_WINDOW_7_SCRATCH_MEM];
const struct dmub_fw_meta_info *fw_info;
uint32_t fw_state_size = DMUB_FW_STATE_SIZE;
uint32_t trace_buffer_size = DMUB_TRACE_BUFFER_SIZE;
uint32_t scratch_mem_size = DMUB_SCRATCH_MEM_SIZE;
if (!dmub->sw_init)
return DMUB_STATUS_INVALID;
@ -256,7 +261,10 @@ dmub_srv_calc_region_info(struct dmub_srv *dmub,
fw_state->base = dmub_align(trace_buff->top, 256);
fw_state->top = fw_state->base + dmub_align(fw_state_size, 64);
out->fb_size = dmub_align(fw_state->top, 4096);
scratch_mem->base = dmub_align(fw_state->top, 256);
scratch_mem->top = scratch_mem->base + dmub_align(scratch_mem_size, 64);
out->fb_size = dmub_align(scratch_mem->top, 4096);
return DMUB_STATUS_OK;
}
@ -334,6 +342,7 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub,
struct dmub_fb *mail_fb = params->fb[DMUB_WINDOW_4_MAILBOX];
struct dmub_fb *tracebuff_fb = params->fb[DMUB_WINDOW_5_TRACEBUFF];
struct dmub_fb *fw_state_fb = params->fb[DMUB_WINDOW_6_FW_STATE];
struct dmub_fb *scratch_mem_fb = params->fb[DMUB_WINDOW_7_SCRATCH_MEM];
struct dmub_rb_init_params rb_params;
struct dmub_window cw0, cw1, cw2, cw3, cw4, cw5, cw6;
@ -370,7 +379,7 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub,
dmub->hw_funcs.reset(dmub);
if (inst_fb && data_fb && bios_fb && mail_fb && tracebuff_fb &&
fw_state_fb) {
fw_state_fb && scratch_mem_fb) {
cw2.offset.quad_part = data_fb->gpu_addr;
cw2.region.base = DMUB_CW0_BASE + inst_fb->size;
cw2.region.top = cw2.region.base + data_fb->size;
@ -396,6 +405,8 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub,
dmub->fw_state = fw_state_fb->cpu_addr;
dmub->scratch_mem_fb = *scratch_mem_fb;
if (dmub->hw_funcs.setup_windows)
dmub->hw_funcs.setup_windows(dmub, &cw2, &cw3, &cw4,
&cw5, &cw6);

5
drivers/gpu/drm/amd/display/modules/hdcp/hdcp.c
View File

@ -328,7 +328,8 @@ enum mod_hdcp_status mod_hdcp_add_display(struct mod_hdcp *hdcp,
/* add display to connection */
hdcp->connection.link = *link;
*display_container = *display;
status = mod_hdcp_add_display_to_topology(hdcp, display->index);
status = mod_hdcp_add_display_to_topology(hdcp, display_container);
if (status != MOD_HDCP_STATUS_SUCCESS)
goto out;
@ -374,7 +375,7 @@ enum mod_hdcp_status mod_hdcp_remove_display(struct mod_hdcp *hdcp,
status = mod_hdcp_remove_display_from_topology(hdcp, index);
if (status != MOD_HDCP_STATUS_SUCCESS)
goto out;
display->state = MOD_HDCP_DISPLAY_INACTIVE;
memset(display, 0, sizeof(struct mod_hdcp_display));
/* request authentication when connection is not reset */
if (current_state(hdcp) != HDCP_UNINITIALIZED)

28
drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h
View File

@ -328,7 +328,7 @@ void mod_hdcp_dump_binary_message(uint8_t *msg, uint32_t msg_size,
/* psp functions */
enum mod_hdcp_status mod_hdcp_add_display_to_topology(
struct mod_hdcp *hdcp, uint8_t index);
struct mod_hdcp *hdcp, struct mod_hdcp_display *display);
enum mod_hdcp_status mod_hdcp_remove_display_from_topology(
struct mod_hdcp *hdcp, uint8_t index);
enum mod_hdcp_status mod_hdcp_hdcp1_create_session(struct mod_hdcp *hdcp);
@ -503,11 +503,6 @@ static inline uint8_t is_display_active(struct mod_hdcp_display *display)
return display->state >= MOD_HDCP_DISPLAY_ACTIVE;
}
static inline uint8_t is_display_added(struct mod_hdcp_display *display)
{
return display->state >= MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED;
}
static inline uint8_t is_display_encryption_enabled(struct mod_hdcp_display *display)
{
return display->state >= MOD_HDCP_DISPLAY_ENCRYPTION_ENABLED;
@ -515,34 +510,23 @@ static inline uint8_t is_display_encryption_enabled(struct mod_hdcp_display *dis
static inline uint8_t get_active_display_count(struct mod_hdcp *hdcp)
{
uint8_t added_count = 0;
uint8_t active_count = 0;
uint8_t i;
for (i = 0; i < MAX_NUM_OF_DISPLAYS; i++)
if (is_display_active(&hdcp->displays[i]))
added_count++;
return added_count;
active_count++;
return active_count;
}
static inline uint8_t get_added_display_count(struct mod_hdcp *hdcp)
{
uint8_t added_count = 0;
uint8_t i;
for (i = 0; i < MAX_NUM_OF_DISPLAYS; i++)
if (is_display_added(&hdcp->displays[i]))
added_count++;
return added_count;
}
static inline struct mod_hdcp_display *get_first_added_display(
static inline struct mod_hdcp_display *get_first_active_display(
struct mod_hdcp *hdcp)
{
uint8_t i;
struct mod_hdcp_display *display = NULL;
for (i = 0; i < MAX_NUM_OF_DISPLAYS; i++)
if (is_display_added(&hdcp->displays[i])) {
if (is_display_active(&hdcp->displays[i])) {
display = &hdcp->displays[i];
break;
}

2
drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c
View File

@ -129,7 +129,7 @@ static inline uint8_t get_device_count(struct mod_hdcp *hdcp)
static inline enum mod_hdcp_status check_device_count(struct mod_hdcp *hdcp)
{
/* device count must be greater than or equal to tracked hdcp displays */
return (get_device_count(hdcp) < get_added_display_count(hdcp)) ?
return (get_device_count(hdcp) < get_active_display_count(hdcp)) ?
MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE :
MOD_HDCP_STATUS_SUCCESS;
}

2
drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c
View File

@ -208,7 +208,7 @@ static inline uint8_t get_device_count(struct mod_hdcp *hdcp)
static enum mod_hdcp_status check_device_count(struct mod_hdcp *hdcp)
{
/* device count must be greater than or equal to tracked hdcp displays */
return (get_device_count(hdcp) < get_added_display_count(hdcp)) ?
return (get_device_count(hdcp) < get_active_display_count(hdcp)) ?
MOD_HDCP_STATUS_HDCP2_DEVICE_COUNT_MISMATCH_FAILURE :
MOD_HDCP_STATUS_SUCCESS;
}

12
drivers/gpu/drm/amd/display/modules/hdcp/hdcp_ddc.c
View File

@ -408,7 +408,7 @@ enum mod_hdcp_status mod_hdcp_read_ake_cert(struct mod_hdcp *hdcp)
enum mod_hdcp_status status;
if (is_dp_hdcp(hdcp)) {
hdcp->auth.msg.hdcp2.ake_cert[0] = 3;
hdcp->auth.msg.hdcp2.ake_cert[0] = HDCP_2_2_AKE_SEND_CERT;
status = read(hdcp, MOD_HDCP_MESSAGE_ID_READ_AKE_SEND_CERT,
hdcp->auth.msg.hdcp2.ake_cert+1,
sizeof(hdcp->auth.msg.hdcp2.ake_cert)-1);
@ -426,7 +426,7 @@ enum mod_hdcp_status mod_hdcp_read_h_prime(struct mod_hdcp *hdcp)
enum mod_hdcp_status status;
if (is_dp_hdcp(hdcp)) {
hdcp->auth.msg.hdcp2.ake_h_prime[0] = 7;
hdcp->auth.msg.hdcp2.ake_h_prime[0] = HDCP_2_2_AKE_SEND_HPRIME;
status = read(hdcp, MOD_HDCP_MESSAGE_ID_READ_AKE_SEND_H_PRIME,
hdcp->auth.msg.hdcp2.ake_h_prime+1,
sizeof(hdcp->auth.msg.hdcp2.ake_h_prime)-1);
@ -444,7 +444,7 @@ enum mod_hdcp_status mod_hdcp_read_pairing_info(struct mod_hdcp *hdcp)
enum mod_hdcp_status status;
if (is_dp_hdcp(hdcp)) {
hdcp->auth.msg.hdcp2.ake_pairing_info[0] = 8;
hdcp->auth.msg.hdcp2.ake_pairing_info[0] = HDCP_2_2_AKE_SEND_PAIRING_INFO;
status = read(hdcp, MOD_HDCP_MESSAGE_ID_READ_AKE_SEND_PAIRING_INFO,
hdcp->auth.msg.hdcp2.ake_pairing_info+1,
sizeof(hdcp->auth.msg.hdcp2.ake_pairing_info)-1);
@ -462,7 +462,7 @@ enum mod_hdcp_status mod_hdcp_read_l_prime(struct mod_hdcp *hdcp)
enum mod_hdcp_status status;
if (is_dp_hdcp(hdcp)) {
hdcp->auth.msg.hdcp2.lc_l_prime[0] = 10;
hdcp->auth.msg.hdcp2.lc_l_prime[0] = HDCP_2_2_LC_SEND_LPRIME;
status = read(hdcp, MOD_HDCP_MESSAGE_ID_READ_LC_SEND_L_PRIME,
hdcp->auth.msg.hdcp2.lc_l_prime+1,
sizeof(hdcp->auth.msg.hdcp2.lc_l_prime)-1);
@ -484,7 +484,7 @@ enum mod_hdcp_status mod_hdcp_read_rx_id_list(struct mod_hdcp *hdcp)
uint32_t rx_id_list_size = 0;
uint32_t bytes_read = 0;
hdcp->auth.msg.hdcp2.rx_id_list[0] = 12;
hdcp->auth.msg.hdcp2.rx_id_list[0] = HDCP_2_2_REP_SEND_RECVID_LIST;
status = read(hdcp, MOD_HDCP_MESSAGE_ID_READ_REPEATER_AUTH_SEND_RECEIVERID_LIST,
hdcp->auth.msg.hdcp2.rx_id_list+1,
HDCP_MAX_AUX_TRANSACTION_SIZE);
@ -511,7 +511,7 @@ enum mod_hdcp_status mod_hdcp_read_stream_ready(struct mod_hdcp *hdcp)
enum mod_hdcp_status status;
if (is_dp_hdcp(hdcp)) {
hdcp->auth.msg.hdcp2.repeater_auth_stream_ready[0] = 17;
hdcp->auth.msg.hdcp2.repeater_auth_stream_ready[0] = HDCP_2_2_REP_STREAM_READY;
status = read(hdcp, MOD_HDCP_MESSAGE_ID_READ_REPEATER_AUTH_STREAM_READY,
hdcp->auth.msg.hdcp2.repeater_auth_stream_ready+1,
sizeof(hdcp->auth.msg.hdcp2.repeater_auth_stream_ready)-1);

37
drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.c
View File

@ -54,7 +54,7 @@ enum mod_hdcp_status mod_hdcp_remove_display_from_topology(
dtm_cmd = (struct ta_dtm_shared_memory *)psp->dtm_context.dtm_shared_buf;
if (!display || !is_display_added(display))
if (!display || !is_display_active(display))
return MOD_HDCP_STATUS_DISPLAY_NOT_FOUND;
memset(dtm_cmd, 0, sizeof(struct ta_dtm_shared_memory));
@ -76,22 +76,18 @@ enum mod_hdcp_status mod_hdcp_remove_display_from_topology(
}
enum mod_hdcp_status mod_hdcp_add_display_to_topology(
struct mod_hdcp *hdcp, uint8_t index)
struct mod_hdcp *hdcp, struct mod_hdcp_display *display)
{
struct psp_context *psp = hdcp->config.psp.handle;
struct ta_dtm_shared_memory *dtm_cmd;
struct mod_hdcp_display *display =
get_active_display_at_index(hdcp, index);
struct mod_hdcp_link *link = &hdcp->connection.link;
if (!psp->dtm_context.dtm_initialized) {
DRM_ERROR("Failed to add display topology, DTM TA is not initialized.");
display->state = MOD_HDCP_DISPLAY_INACTIVE;
return MOD_HDCP_STATUS_FAILURE;
}
if (!display || is_display_added(display))
return MOD_HDCP_STATUS_UPDATE_TOPOLOGY_FAILURE;
dtm_cmd = (struct ta_dtm_shared_memory *)psp->dtm_context.dtm_shared_buf;
memset(dtm_cmd, 0, sizeof(struct ta_dtm_shared_memory));
@ -113,20 +109,21 @@ enum mod_hdcp_status mod_hdcp_add_display_to_topology(
psp_dtm_invoke(psp, dtm_cmd->cmd_id);
if (dtm_cmd->dtm_status != TA_DTM_STATUS__SUCCESS)
if (dtm_cmd->dtm_status != TA_DTM_STATUS__SUCCESS) {
display->state = MOD_HDCP_DISPLAY_INACTIVE;
return MOD_HDCP_STATUS_UPDATE_TOPOLOGY_FAILURE;
}
display->state = MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED;
HDCP_TOP_ADD_DISPLAY_TRACE(hdcp, display->index);
return MOD_HDCP_STATUS_SUCCESS;
return MOD_HDCP_STATUS_SUCCESS;
}
enum mod_hdcp_status mod_hdcp_hdcp1_create_session(struct mod_hdcp *hdcp)
{
struct psp_context *psp = hdcp->config.psp.handle;
struct mod_hdcp_display *display = get_first_added_display(hdcp);
struct mod_hdcp_display *display = get_first_active_display(hdcp);
struct ta_hdcp_shared_memory *hdcp_cmd;
if (!psp->hdcp_context.hdcp_initialized) {
@ -179,7 +176,7 @@ enum mod_hdcp_status mod_hdcp_hdcp1_destroy_session(struct mod_hdcp *hdcp)
if (is_display_encryption_enabled(
&hdcp->displays[i])) {
hdcp->displays[i].state =
MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED;
MOD_HDCP_DISPLAY_ACTIVE;
HDCP_HDCP1_DISABLED_TRACE(hdcp,
hdcp->displays[i].index);
}
@ -231,7 +228,7 @@ enum mod_hdcp_status mod_hdcp_hdcp1_enable_encryption(struct mod_hdcp *hdcp)
{
struct psp_context *psp = hdcp->config.psp.handle;
struct ta_hdcp_shared_memory *hdcp_cmd;
struct mod_hdcp_display *display = get_first_added_display(hdcp);
struct mod_hdcp_display *display = get_first_active_display(hdcp);
hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.hdcp_shared_buf;
memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory));
@ -301,8 +298,7 @@ enum mod_hdcp_status mod_hdcp_hdcp1_enable_dp_stream_encryption(struct mod_hdcp
for (i = 0; i < MAX_NUM_OF_DISPLAYS; i++) {
if (hdcp->displays[i].state != MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED ||
hdcp->displays[i].adjust.disable)
if (hdcp->displays[i].adjust.disable)
continue;
memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory));
@ -364,7 +360,7 @@ enum mod_hdcp_status mod_hdcp_hdcp2_create_session(struct mod_hdcp *hdcp)
{
struct psp_context *psp = hdcp->config.psp.handle;
struct ta_hdcp_shared_memory *hdcp_cmd;
struct mod_hdcp_display *display = get_first_added_display(hdcp);
struct mod_hdcp_display *display = get_first_active_display(hdcp);
if (!psp->hdcp_context.hdcp_initialized) {
DRM_ERROR("Failed to create hdcp session, HDCP TA is not initialized");
@ -423,7 +419,7 @@ enum mod_hdcp_status mod_hdcp_hdcp2_destroy_session(struct mod_hdcp *hdcp)
if (is_display_encryption_enabled(
&hdcp->displays[i])) {
hdcp->displays[i].state =
MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED;
MOD_HDCP_DISPLAY_ACTIVE;
HDCP_HDCP2_DISABLED_TRACE(hdcp,
hdcp->displays[i].index);
}
@ -662,7 +658,7 @@ enum mod_hdcp_status mod_hdcp_hdcp2_enable_encryption(struct mod_hdcp *hdcp)
{
struct psp_context *psp = hdcp->config.psp.handle;
struct ta_hdcp_shared_memory *hdcp_cmd;
struct mod_hdcp_display *display = get_first_added_display(hdcp);
struct mod_hdcp_display *display = get_first_active_display(hdcp);
hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.hdcp_shared_buf;
memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory));
@ -747,8 +743,7 @@ enum mod_hdcp_status mod_hdcp_hdcp2_enable_dp_stream_encryption(struct mod_hdcp
for (i = 0; i < MAX_NUM_OF_DISPLAYS; i++) {
if (hdcp->displays[i].state != MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED ||
hdcp->displays[i].adjust.disable)
if (hdcp->displays[i].adjust.disable)
continue;
hdcp_cmd->in_msg.hdcp2_enable_dp_stream_encryption.display_handle = hdcp->displays[i].index;
hdcp_cmd->in_msg.hdcp2_enable_dp_stream_encryption.session_handle = hdcp->auth.id;

1
drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h
View File

@ -117,7 +117,6 @@ enum mod_hdcp_operation_mode {
enum mod_hdcp_display_state {
MOD_HDCP_DISPLAY_INACTIVE = 0,
MOD_HDCP_DISPLAY_ACTIVE,
MOD_HDCP_DISPLAY_ACTIVE_AND_ADDED,
MOD_HDCP_DISPLAY_ENCRYPTION_ENABLED
};

3
drivers/gpu/drm/amd/display/modules/inc/mod_info_packet.h
View File

@ -34,8 +34,7 @@ struct dc_info_packet;
struct mod_vrr_params;
void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
struct dc_info_packet *info_packet,
bool *use_vsc_sdp_for_colorimetry);
struct dc_info_packet *info_packet);
void mod_build_hf_vsif_infopacket(const struct dc_stream_state *stream,
struct dc_info_packet *info_packet, int ALLMEnabled, int ALLMValue);

20
drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c
View File

@ -130,8 +130,7 @@ enum ColorimetryYCCDP {
};
void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
struct dc_info_packet *info_packet,
bool *use_vsc_sdp_for_colorimetry)
struct dc_info_packet *info_packet)
{
unsigned int vsc_packet_revision = vsc_packet_undefined;
unsigned int i;
@ -139,11 +138,6 @@ void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
unsigned int colorimetryFormat = 0;
bool stereo3dSupport = false;
/* Initialize first, later if infopacket is valid determine if VSC SDP
* should be used to signal colorimetry format and pixel encoding.
*/
*use_vsc_sdp_for_colorimetry = false;
if (stream->timing.timing_3d_format != TIMING_3D_FORMAT_NONE && stream->view_format != VIEW_3D_FORMAT_NONE) {
vsc_packet_revision = vsc_packet_rev1;
stereo3dSupport = true;
@ -153,9 +147,8 @@ void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
if (stream->psr_version != 0)
vsc_packet_revision = vsc_packet_rev2;
/* Update to revision 5 for extended colorimetry support for DPCD 1.4+ */
if (stream->link->dpcd_caps.dpcd_rev.raw >= 0x14 &&
stream->link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED)
/* Update to revision 5 for extended colorimetry support */
if (stream->use_vsc_sdp_for_colorimetry)
vsc_packet_revision = vsc_packet_rev5;
/* VSC packet not needed based on the features
@ -269,13 +262,6 @@ void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
info_packet->valid = true;
/* If we are using VSC SDP revision 05h, use this to signal for
* colorimetry format and pixel encoding. HW should later be
* programmed to set MSA MISC1 bit 6 to indicate ignore
* colorimetry format and pixel encoding in the MSA.
*/
*use_vsc_sdp_for_colorimetry = true;
/* Set VSC SDP fields for pixel encoding and colorimetry format from DP 1.3 specs
* Data Bytes DB 18~16
* Bits 3:0 (Colorimetry Format) | Bits 7:4 (Pixel Encoding)

34
drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
View File

@ -23,15 +23,12 @@
#include <linux/firmware.h>
#include <linux/pci.h>
#include "pp_debug.h"
#include "amdgpu.h"
#include "amdgpu_smu.h"
#include "smu_internal.h"
#include "soc15_common.h"
#include "smu_v11_0.h"
#include "smu_v12_0.h"
#include "atom.h"
#include "amd_pcie.h"
#include "vega20_ppt.h"
#include "arcturus_ppt.h"
#include "navi10_ppt.h"
@ -935,6 +932,13 @@ static int smu_sw_init(void *handle)
return ret;
}
if (adev->smu.ppt_funcs->i2c_eeprom_init) {
ret = smu_i2c_eeprom_init(smu, &adev->pm.smu_i2c);
if (ret)
return ret;
}
return 0;
}
@ -944,6 +948,9 @@ static int smu_sw_fini(void *handle)
struct smu_context *smu = &adev->smu;
int ret;
if (adev->smu.ppt_funcs->i2c_eeprom_fini)
smu_i2c_eeprom_fini(smu, &adev->pm.smu_i2c);
kfree(smu->irq_source);
smu->irq_source = NULL;
@ -1463,21 +1470,26 @@ static int smu_disable_dpm(struct smu_context *smu)
}
/*
* For baco on Arcturus, this operation
* (disable all smu feature) will be handled by SMU FW.
* Disable all enabled SMU features.
* This should be handled in SMU FW, as a backup
* driver can issue call to SMU FW until sequence
* in SMU FW is operational.
*/
if (adev->asic_type == CHIP_ARCTURUS) {
if (use_baco && (smu_version > 0x360e00))
return 0;
}
/* Disable all enabled SMU features */
ret = smu_system_features_control(smu, false);
if (ret) {
pr_err("Failed to disable smu features.\n");
return ret;
}
/*
* Arcturus does not have BACO bit in disable feature mask.
* Enablement of BACO bit on Arcturus should be skipped.
*/
if (adev->asic_type == CHIP_ARCTURUS) {
if (use_baco && (smu_version > 0x360e00))
return 0;
}
/* For baco, need to leave BACO feature enabled */
if (use_baco) {
/*

5
drivers/gpu/drm/amd/powerplay/arcturus_ppt.c
View File

@ -21,7 +21,6 @@
*
*/
#include "pp_debug.h"
#include <linux/firmware.h>
#include "amdgpu.h"
#include "amdgpu_smu.h"
@ -42,7 +41,7 @@
#include <linux/pci.h>
#include "amdgpu_ras.h"
#define to_amdgpu_device(x) (container_of(x, struct amdgpu_ras, eeprom_control.eeprom_accessor))->adev
#define to_amdgpu_device(x) (container_of(x, struct amdgpu_device, pm.smu_i2c))
#define CTF_OFFSET_EDGE 5
#define CTF_OFFSET_HOTSPOT 5
@ -2191,7 +2190,7 @@ static int arcturus_i2c_eeprom_control_init(struct i2c_adapter *control)
control->class = I2C_CLASS_SPD;
control->dev.parent = &adev->pdev->dev;
control->algo = &arcturus_i2c_eeprom_i2c_algo;
snprintf(control->name, sizeof(control->name), "RAS EEPROM");
snprintf(control->name, sizeof(control->name), "AMDGPU EEPROM");
res = i2c_add_adapter(control);
if (res)

6
drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c
View File

@ -1250,7 +1250,7 @@ static void smu7_set_dpm_event_sources(struct pp_hwmgr *hwmgr, uint32_t sources)
switch (sources) {
default:
pr_err("Unknown throttling event sources.");
/* fall through */
fallthrough;
case 0:
protection = false;
/* src is unused */
@ -3698,12 +3698,12 @@ static int smu7_request_link_speed_change_before_state_change(
data->force_pcie_gen = PP_PCIEGen2;
if (current_link_speed == PP_PCIEGen2)
break;
/* fall through */
fallthrough;
case PP_PCIEGen2:
if (0 == amdgpu_acpi_pcie_performance_request(hwmgr->adev, PCIE_PERF_REQ_GEN2, false))
break;
fallthrough;
#endif
/* fall through */
default:
data->force_pcie_gen = smu7_get_current_pcie_speed(hwmgr);
break;

2
drivers/gpu/drm/amd/powerplay/navi10_ppt.c
View File

@ -21,7 +21,6 @@
*
*/
#include "pp_debug.h"
#include <linux/firmware.h>
#include <linux/pci.h>
#include "amdgpu.h"
@ -31,7 +30,6 @@
#include "amdgpu_atomfirmware.h"
#include "smu_v11_0.h"
#include "smu11_driver_if_navi10.h"
#include "soc15_common.h"
#include "atom.h"
#include "navi10_ppt.h"
#include "smu_v11_0_pptable.h"

1
drivers/gpu/drm/amd/powerplay/renoir_ppt.c
View File

@ -24,7 +24,6 @@
#include "amdgpu.h"
#include "amdgpu_smu.h"
#include "smu_internal.h"
#include "soc15_common.h"
#include "smu_v12_0_ppsmc.h"
#include "smu12_driver_if.h"
#include "smu_v12_0.h"

1
drivers/gpu/drm/amd/powerplay/smu_v11_0.c
View File

@ -26,7 +26,6 @@
#define SMU_11_0_PARTIAL_PPTABLE
#include "pp_debug.h"
#include "amdgpu.h"
#include "amdgpu_smu.h"
#include "smu_internal.h"

1
drivers/gpu/drm/amd/powerplay/smu_v12_0.c
View File

@ -20,7 +20,6 @@
* OTHER DEALINGS IN THE SOFTWARE.
*/
#include "pp_debug.h"
#include <linux/firmware.h>
#include "amdgpu.h"
#include "amdgpu_smu.h"

11
drivers/gpu/drm/amd/powerplay/smumgr/vega20_smumgr.c
View File

@ -33,6 +33,8 @@
#include "smu7_smumgr.h"
#include "vega20_hwmgr.h"
#include "smu_v11_0_i2c.h"
/* MP Apertures */
#define MP0_Public 0x03800000
#define MP0_SRAM 0x03900000
@ -406,6 +408,7 @@ static int vega20_smu_init(struct pp_hwmgr *hwmgr)
struct vega20_smumgr *priv;
unsigned long tools_size = 0x19000;
int ret = 0;
struct amdgpu_device *adev = hwmgr->adev;
struct cgs_firmware_info info = {0};
@ -505,6 +508,10 @@ static int vega20_smu_init(struct pp_hwmgr *hwmgr)
priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].version = 0x01;
priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].size = sizeof(DpmActivityMonitorCoeffInt_t);
ret = smu_v11_0_i2c_eeprom_control_init(&adev->pm.smu_i2c);
if (ret)
goto err4;
return 0;
err4:
@ -537,6 +544,9 @@ static int vega20_smu_fini(struct pp_hwmgr *hwmgr)
{
struct vega20_smumgr *priv =
(struct vega20_smumgr *)(hwmgr->smu_backend);
struct amdgpu_device *adev = hwmgr->adev;
smu_v11_0_i2c_eeprom_control_fini(&adev->pm.smu_i2c);
if (priv) {
amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_PPTABLE].handle,
@ -560,6 +570,7 @@ static int vega20_smu_fini(struct pp_hwmgr *hwmgr)
kfree(hwmgr->smu_backend);
hwmgr->smu_backend = NULL;
}
return 0;
}

1
drivers/gpu/drm/amd/powerplay/vega20_ppt.c
View File

@ -21,7 +21,6 @@
*
*/
#include "pp_debug.h"
#include <linux/firmware.h>
#include "amdgpu.h"
#include "amdgpu_smu.h"

27
drivers/gpu/drm/scheduler/gpu_scheduler_trace.h
View File

@ -59,6 +59,33 @@ TRACE_EVENT(drm_sched_job,
__entry->job_count, __entry->hw_job_count)
);
TRACE_EVENT(drm_run_job,
TP_PROTO(struct drm_sched_job *sched_job, struct drm_sched_entity *entity),
TP_ARGS(sched_job, entity),
TP_STRUCT__entry(
__field(struct drm_sched_entity *, entity)
__field(struct dma_fence *, fence)
__field(const char *, name)
__field(uint64_t, id)
__field(u32, job_count)
__field(int, hw_job_count)
),
TP_fast_assign(
__entry->entity = entity;
__entry->id = sched_job->id;
__entry->fence = &sched_job->s_fence->finished;
__entry->name = sched_job->sched->name;
__entry->job_count = spsc_queue_count(&entity->job_queue);
__entry->hw_job_count = atomic_read(
&sched_job->sched->hw_rq_count);
),
TP_printk("entity=%p, id=%llu, fence=%p, ring=%s, job count:%u, hw job count:%d",
__entry->entity, __entry->id,
__entry->fence, __entry->name,
__entry->job_count, __entry->hw_job_count)
);
TRACE_EVENT(drm_sched_process_job,
TP_PROTO(struct drm_sched_fence *fence),
TP_ARGS(fence),

38
drivers/gpu/drm/scheduler/sched_entity.c
View File

@ -137,38 +137,6 @@ bool drm_sched_entity_is_ready(struct drm_sched_entity *entity)
return true;
}
/**
* drm_sched_entity_get_free_sched - Get the rq from rq_list with least load
*
* @entity: scheduler entity
*
* Return the pointer to the rq with least load.
*/
static struct drm_sched_rq *
drm_sched_entity_get_free_sched(struct drm_sched_entity *entity)
{
struct drm_sched_rq *rq = NULL;
unsigned int min_score = UINT_MAX, num_score;
int i;
for (i = 0; i < entity->num_sched_list; ++i) {
struct drm_gpu_scheduler *sched = entity->sched_list[i];
if (!entity->sched_list[i]->ready) {
DRM_WARN("sched%s is not ready, skipping", sched->name);
continue;
}
num_score = atomic_read(&sched->score);
if (num_score < min_score) {
min_score = num_score;
rq = &entity->sched_list[i]->sched_rq[entity->priority];
}
}
return rq;
}
/**
* drm_sched_entity_flush - Flush a context entity
*
@ -479,6 +447,7 @@ struct drm_sched_job *drm_sched_entity_pop_job(struct drm_sched_entity *entity)
void drm_sched_entity_select_rq(struct drm_sched_entity *entity)
{
struct dma_fence *fence;
struct drm_gpu_scheduler *sched;
struct drm_sched_rq *rq;
if (spsc_queue_count(&entity->job_queue) || entity->num_sched_list <= 1)
@ -489,7 +458,8 @@ void drm_sched_entity_select_rq(struct drm_sched_entity *entity)
return;
spin_lock(&entity->rq_lock);
rq = drm_sched_entity_get_free_sched(entity);
sched = drm_sched_pick_best(entity->sched_list, entity->num_sched_list);
rq = sched ? &sched->sched_rq[entity->priority] : NULL;
if (rq != entity->rq) {
drm_sched_rq_remove_entity(entity->rq, entity);
entity->rq = rq;
@ -516,7 +486,7 @@ void drm_sched_entity_push_job(struct drm_sched_job *sched_job,
bool first;
trace_drm_sched_job(sched_job, entity);
atomic_inc(&entity->rq->sched->score);
atomic_inc(&entity->rq->sched->num_jobs);
WRITE_ONCE(entity->last_user, current->group_leader);
first = spsc_queue_push(&entity->job_queue, &sched_job->queue_node);

86
drivers/gpu/drm/scheduler/sched_main.c
View File

@ -92,7 +92,6 @@ void drm_sched_rq_add_entity(struct drm_sched_rq *rq,
if (!list_empty(&entity->list))
return;
spin_lock(&rq->lock);
atomic_inc(&rq->sched->score);
list_add_tail(&entity->list, &rq->entities);
spin_unlock(&rq->lock);
}
@ -111,7 +110,6 @@ void drm_sched_rq_remove_entity(struct drm_sched_rq *rq,
if (list_empty(&entity->list))
return;
spin_lock(&rq->lock);
atomic_dec(&rq->sched->score);
list_del_init(&entity->list);
if (rq->current_entity == entity)
rq->current_entity = NULL;
@ -222,8 +220,7 @@ EXPORT_SYMBOL(drm_sched_fault);
*
* Suspend the delayed work timeout for the scheduler. This is done by
* modifying the delayed work timeout to an arbitrary large value,
* MAX_SCHEDULE_TIMEOUT in this case. Note that this function can be
* called from an IRQ context.
* MAX_SCHEDULE_TIMEOUT in this case.
*
* Returns the timeout remaining
*
@ -252,46 +249,41 @@ EXPORT_SYMBOL(drm_sched_suspend_timeout);
* @sched: scheduler instance for which to resume the timeout
* @remaining: remaining timeout
*
* Resume the delayed work timeout for the scheduler. Note that
* this function can be called from an IRQ context.
* Resume the delayed work timeout for the scheduler.
*/
void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched,
unsigned long remaining)
{
unsigned long flags;
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
if (list_empty(&sched->ring_mirror_list))
cancel_delayed_work(&sched->work_tdr);
else
mod_delayed_work(system_wq, &sched->work_tdr, remaining);
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
}
EXPORT_SYMBOL(drm_sched_resume_timeout);
static void drm_sched_job_begin(struct drm_sched_job *s_job)
{
struct drm_gpu_scheduler *sched = s_job->sched;
unsigned long flags;
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
list_add_tail(&s_job->node, &sched->ring_mirror_list);
drm_sched_start_timeout(sched);
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
}
static void drm_sched_job_timedout(struct work_struct *work)
{
struct drm_gpu_scheduler *sched;
struct drm_sched_job *job;
unsigned long flags;
sched = container_of(work, struct drm_gpu_scheduler, work_tdr.work);
/* Protects against concurrent deletion in drm_sched_get_cleanup_job */
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
job = list_first_entry_or_null(&sched->ring_mirror_list,
struct drm_sched_job, node);
@ -302,7 +294,7 @@ static void drm_sched_job_timedout(struct work_struct *work)
* is parked at which point it's safe.
*/
list_del_init(&job->node);
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
job->sched->ops->timedout_job(job);
@ -315,12 +307,12 @@ static void drm_sched_job_timedout(struct work_struct *work)
sched->free_guilty = false;
}
} else {
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
}
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
drm_sched_start_timeout(sched);
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
}
/**
@ -383,7 +375,6 @@ EXPORT_SYMBOL(drm_sched_increase_karma);
void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad)
{
struct drm_sched_job *s_job, *tmp;
unsigned long flags;
kthread_park(sched->thread);
@ -417,9 +408,9 @@ void drm_sched_stop(struct drm_gpu_scheduler *sched, struct drm_sched_job *bad)
* remove job from ring_mirror_list.
* Locking here is for concurrent resume timeout
*/
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
list_del_init(&s_job->node);
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
/*
* Wait for job's HW fence callback to finish using s_job
@ -462,7 +453,6 @@ EXPORT_SYMBOL(drm_sched_stop);
void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery)
{
struct drm_sched_job *s_job, *tmp;
unsigned long flags;
int r;
/*
@ -491,9 +481,9 @@ void drm_sched_start(struct drm_gpu_scheduler *sched, bool full_recovery)
}
if (full_recovery) {
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
drm_sched_start_timeout(sched);
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
}
kthread_unpark(sched->thread);
@ -657,7 +647,7 @@ static void drm_sched_process_job(struct dma_fence *f, struct dma_fence_cb *cb)
struct drm_gpu_scheduler *sched = s_fence->sched;
atomic_dec(&sched->hw_rq_count);
atomic_dec(&sched->score);
atomic_dec(&sched->num_jobs);
trace_drm_sched_process_job(s_fence);
@ -677,7 +667,6 @@ static struct drm_sched_job *
drm_sched_get_cleanup_job(struct drm_gpu_scheduler *sched)
{
struct drm_sched_job *job;
unsigned long flags;
/*
* Don't destroy jobs while the timeout worker is running OR thread
@ -688,7 +677,7 @@ drm_sched_get_cleanup_job(struct drm_gpu_scheduler *sched)
__kthread_should_park(sched->thread))
return NULL;
spin_lock_irqsave(&sched->job_list_lock, flags);
spin_lock(&sched->job_list_lock);
job = list_first_entry_or_null(&sched->ring_mirror_list,
struct drm_sched_job, node);
@ -702,11 +691,47 @@ drm_sched_get_cleanup_job(struct drm_gpu_scheduler *sched)
drm_sched_start_timeout(sched);
}
spin_unlock_irqrestore(&sched->job_list_lock, flags);
spin_unlock(&sched->job_list_lock);
return job;
}
/**
* drm_sched_pick_best - Get a drm sched from a sched_list with the least load
* @sched_list: list of drm_gpu_schedulers
* @num_sched_list: number of drm_gpu_schedulers in the sched_list
*
* Returns pointer of the sched with the least load or NULL if none of the
* drm_gpu_schedulers are ready
*/
struct drm_gpu_scheduler *
drm_sched_pick_best(struct drm_gpu_scheduler **sched_list,
unsigned int num_sched_list)
{
struct drm_gpu_scheduler *sched, *picked_sched = NULL;
int i;
unsigned int min_jobs = UINT_MAX, num_jobs;
for (i = 0; i < num_sched_list; ++i) {
sched = sched_list[i];
if (!sched->ready) {
DRM_WARN("scheduler %s is not ready, skipping",
sched->name);
continue;
}
num_jobs = atomic_read(&sched->num_jobs);
if (num_jobs < min_jobs) {
min_jobs = num_jobs;
picked_sched = sched;
}
}
return picked_sched;
}
EXPORT_SYMBOL(drm_sched_pick_best);
/**
* drm_sched_blocked - check if the scheduler is blocked
*
@ -773,6 +798,7 @@ static int drm_sched_main(void *param)
atomic_inc(&sched->hw_rq_count);
drm_sched_job_begin(sched_job);
trace_drm_run_job(sched_job, entity);
fence = sched->ops->run_job(sched_job);
drm_sched_fence_scheduled(s_fence);
@ -832,7 +858,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched,
spin_lock_init(&sched->job_list_lock);
atomic_set(&sched->hw_rq_count, 0);
INIT_DELAYED_WORK(&sched->work_tdr, drm_sched_job_timedout);
atomic_set(&sched->score, 0);
atomic_set(&sched->num_jobs, 0);
atomic64_set(&sched->job_id_count, 0);
/* Each scheduler will run on a seperate kernel thread */

9
include/drm/gpu_scheduler.h
View File

@ -262,7 +262,7 @@ struct drm_sched_backend_ops {
* @job_list_lock: lock to protect the ring_mirror_list.
* @hang_limit: once the hangs by a job crosses this limit then it is marked
* guilty and it will be considered for scheduling further.
* @score: score to help loadbalancer pick a idle sched
* @num_jobs: the number of jobs in queue in the scheduler
* @ready: marks if the underlying HW is ready to work
* @free_guilty: A hit to time out handler to free the guilty job.
*
@ -283,8 +283,8 @@ struct drm_gpu_scheduler {
struct list_head ring_mirror_list;
spinlock_t job_list_lock;
int hang_limit;
atomic_t score;
bool ready;
atomic_t num_jobs;
bool ready;
bool free_guilty;
};
@ -341,5 +341,8 @@ void drm_sched_fence_finished(struct drm_sched_fence *fence);
unsigned long drm_sched_suspend_timeout(struct drm_gpu_scheduler *sched);
void drm_sched_resume_timeout(struct drm_gpu_scheduler *sched,
unsigned long remaining);
struct drm_gpu_scheduler *
drm_sched_pick_best(struct drm_gpu_scheduler **sched_list,
unsigned int num_sched_list);
#endif