summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu.h4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c19
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_device.c9
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/cz_dpm.c86
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v10_0.c26
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v11_0.c26
-rw-r--r--drivers/gpu/drm/amd/amdgpu/dce_v8_0.c26
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vce_v3_0.c48
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vi.c35
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_process.c9
13 files changed, 238 insertions, 58 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 01657830b470..e9fde72cf038 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -1614,6 +1614,9 @@ struct amdgpu_uvd {
#define AMDGPU_MAX_VCE_HANDLES 16
#define AMDGPU_VCE_FIRMWARE_OFFSET 256
+#define AMDGPU_VCE_HARVEST_VCE0 (1 << 0)
+#define AMDGPU_VCE_HARVEST_VCE1 (1 << 1)
+
struct amdgpu_vce {
struct amdgpu_bo *vcpu_bo;
uint64_t gpu_addr;
@@ -1626,6 +1629,7 @@ struct amdgpu_vce {
const struct firmware *fw; /* VCE firmware */
struct amdgpu_ring ring[AMDGPU_MAX_VCE_RINGS];
struct amdgpu_irq_src irq;
+ unsigned harvest_config;
};
/*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index d63135bf29c0..1f040d85ac47 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -669,6 +669,7 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
struct amdgpu_cs_parser *p)
{
+ struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
struct amdgpu_ib *ib;
int i, j, r;
@@ -694,6 +695,7 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
for (j = 0; j < num_deps; ++j) {
struct amdgpu_fence *fence;
struct amdgpu_ring *ring;
+ struct amdgpu_ctx *ctx;
r = amdgpu_cs_get_ring(adev, deps[j].ip_type,
deps[j].ip_instance,
@@ -701,14 +703,21 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
if (r)
return r;
+ ctx = amdgpu_ctx_get(fpriv, deps[j].ctx_id);
+ if (ctx == NULL)
+ return -EINVAL;
+
r = amdgpu_fence_recreate(ring, p->filp,
deps[j].handle,
&fence);
- if (r)
+ if (r) {
+ amdgpu_ctx_put(ctx);
return r;
+ }
amdgpu_sync_fence(&ib->sync, fence);
amdgpu_fence_unref(&fence);
+ amdgpu_ctx_put(ctx);
}
}
@@ -808,12 +817,16 @@ int amdgpu_cs_wait_ioctl(struct drm_device *dev, void *data,
r = amdgpu_cs_get_ring(adev, wait->in.ip_type, wait->in.ip_instance,
wait->in.ring, &ring);
- if (r)
+ if (r) {
+ amdgpu_ctx_put(ctx);
return r;
+ }
r = amdgpu_fence_recreate(ring, filp, wait->in.handle, &fence);
- if (r)
+ if (r) {
+ amdgpu_ctx_put(ctx);
return r;
+ }
r = fence_wait_timeout(&fence->base, true, timeout);
amdgpu_fence_unref(&fence);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index ba46be361c9b..d79009b65867 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -1207,10 +1207,15 @@ static int amdgpu_early_init(struct amdgpu_device *adev)
} else {
if (adev->ip_blocks[i].funcs->early_init) {
r = adev->ip_blocks[i].funcs->early_init((void *)adev);
- if (r)
+ if (r == -ENOENT)
+ adev->ip_block_enabled[i] = false;
+ else if (r)
return r;
+ else
+ adev->ip_block_enabled[i] = true;
+ } else {
+ adev->ip_block_enabled[i] = true;
}
- adev->ip_block_enabled[i] = true;
}
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 975edb1000a2..ae43b58c9733 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -352,7 +352,7 @@ unsigned long amdgpu_gem_timeout(uint64_t timeout_ns)
if (((int64_t)timeout_ns) < 0)
return MAX_SCHEDULE_TIMEOUT;
- timeout = ktime_sub_ns(ktime_get(), timeout_ns);
+ timeout = ktime_sub(ns_to_ktime(timeout_ns), ktime_get());
if (ktime_to_ns(timeout) < 0)
return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 5533434c7a8f..31ad444c6386 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -459,6 +459,7 @@ static int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file
memcpy(&dev_info.cu_bitmap[0], &cu_info.bitmap[0], sizeof(cu_info.bitmap));
dev_info.vram_type = adev->mc.vram_type;
dev_info.vram_bit_width = adev->mc.vram_width;
+ dev_info.vce_harvest_config = adev->vce.harvest_config;
return copy_to_user(out, &dev_info,
min((size_t)size, sizeof(dev_info))) ? -EFAULT : 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
index f75a31df30bd..ace870afc7d4 100644
--- a/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/cz_dpm.c
@@ -494,29 +494,67 @@ static void cz_dpm_fini(struct amdgpu_device *adev)
amdgpu_free_extended_power_table(adev);
}
+#define ixSMUSVI_NB_CURRENTVID 0xD8230044
+#define CURRENT_NB_VID_MASK 0xff000000
+#define CURRENT_NB_VID__SHIFT 24
+#define ixSMUSVI_GFX_CURRENTVID 0xD8230048
+#define CURRENT_GFX_VID_MASK 0xff000000
+#define CURRENT_GFX_VID__SHIFT 24
+
static void
cz_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev,
struct seq_file *m)
{
+ struct cz_power_info *pi = cz_get_pi(adev);
struct amdgpu_clock_voltage_dependency_table *table =
&adev->pm.dpm.dyn_state.vddc_dependency_on_sclk;
- u32 current_index =
- (RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX) &
- TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX_MASK) >>
- TARGET_AND_CURRENT_PROFILE_INDEX__CURR_SCLK_INDEX__SHIFT;
- u32 sclk, tmp;
- u16 vddc;
-
- if (current_index >= NUM_SCLK_LEVELS) {
- seq_printf(m, "invalid dpm profile %d\n", current_index);
+ struct amdgpu_uvd_clock_voltage_dependency_table *uvd_table =
+ &adev->pm.dpm.dyn_state.uvd_clock_voltage_dependency_table;
+ struct amdgpu_vce_clock_voltage_dependency_table *vce_table =
+ &adev->pm.dpm.dyn_state.vce_clock_voltage_dependency_table;
+ u32 sclk_index = REG_GET_FIELD(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX),
+ TARGET_AND_CURRENT_PROFILE_INDEX, CURR_SCLK_INDEX);
+ u32 uvd_index = REG_GET_FIELD(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_2),
+ TARGET_AND_CURRENT_PROFILE_INDEX_2, CURR_UVD_INDEX);
+ u32 vce_index = REG_GET_FIELD(RREG32_SMC(ixTARGET_AND_CURRENT_PROFILE_INDEX_2),
+ TARGET_AND_CURRENT_PROFILE_INDEX_2, CURR_VCE_INDEX);
+ u32 sclk, vclk, dclk, ecclk, tmp;
+ u16 vddnb, vddgfx;
+
+ if (sclk_index >= NUM_SCLK_LEVELS) {
+ seq_printf(m, "invalid sclk dpm profile %d\n", sclk_index);
} else {
- sclk = table->entries[current_index].clk;
- tmp = (RREG32_SMC(ixSMU_VOLTAGE_STATUS) &
- SMU_VOLTAGE_STATUS__SMU_VOLTAGE_CURRENT_LEVEL_MASK) >>
- SMU_VOLTAGE_STATUS__SMU_VOLTAGE_CURRENT_LEVEL__SHIFT;
- vddc = cz_convert_8bit_index_to_voltage(adev, (u16)tmp);
- seq_printf(m, "power level %d sclk: %u vddc: %u\n",
- current_index, sclk, vddc);
+ sclk = table->entries[sclk_index].clk;
+ seq_printf(m, "%u sclk: %u\n", sclk_index, sclk);
+ }
+
+ tmp = (RREG32_SMC(ixSMUSVI_NB_CURRENTVID) &
+ CURRENT_NB_VID_MASK) >> CURRENT_NB_VID__SHIFT;
+ vddnb = cz_convert_8bit_index_to_voltage(adev, (u16)tmp);
+ tmp = (RREG32_SMC(ixSMUSVI_GFX_CURRENTVID) &
+ CURRENT_GFX_VID_MASK) >> CURRENT_GFX_VID__SHIFT;
+ vddgfx = cz_convert_8bit_index_to_voltage(adev, (u16)tmp);
+ seq_printf(m, "vddnb: %u vddgfx: %u\n", vddnb, vddgfx);
+
+ seq_printf(m, "uvd %sabled\n", pi->uvd_power_gated ? "dis" : "en");
+ if (!pi->uvd_power_gated) {
+ if (uvd_index >= CZ_MAX_HARDWARE_POWERLEVELS) {
+ seq_printf(m, "invalid uvd dpm level %d\n", uvd_index);
+ } else {
+ vclk = uvd_table->entries[uvd_index].vclk;
+ dclk = uvd_table->entries[uvd_index].dclk;
+ seq_printf(m, "%u uvd vclk: %u dclk: %u\n", uvd_index, vclk, dclk);
+ }
+ }
+
+ seq_printf(m, "vce %sabled\n", pi->vce_power_gated ? "dis" : "en");
+ if (!pi->vce_power_gated) {
+ if (vce_index >= CZ_MAX_HARDWARE_POWERLEVELS) {
+ seq_printf(m, "invalid vce dpm level %d\n", vce_index);
+ } else {
+ ecclk = vce_table->entries[vce_index].ecclk;
+ seq_printf(m, "%u vce ecclk: %u\n", vce_index, ecclk);
+ }
}
}
@@ -1679,25 +1717,31 @@ static int cz_dpm_unforce_dpm_levels(struct amdgpu_device *adev)
if (ret)
return ret;
- DRM_INFO("DPM unforce state min=%d, max=%d.\n",
- pi->sclk_dpm.soft_min_clk,
- pi->sclk_dpm.soft_max_clk);
+ DRM_DEBUG("DPM unforce state min=%d, max=%d.\n",
+ pi->sclk_dpm.soft_min_clk,
+ pi->sclk_dpm.soft_max_clk);
return 0;
}
static int cz_dpm_force_dpm_level(struct amdgpu_device *adev,
- enum amdgpu_dpm_forced_level level)
+ enum amdgpu_dpm_forced_level level)
{
int ret = 0;
switch (level) {
case AMDGPU_DPM_FORCED_LEVEL_HIGH:
+ ret = cz_dpm_unforce_dpm_levels(adev);
+ if (ret)
+ return ret;
ret = cz_dpm_force_highest(adev);
if (ret)
return ret;
break;
case AMDGPU_DPM_FORCED_LEVEL_LOW:
+ ret = cz_dpm_unforce_dpm_levels(adev);
+ if (ret)
+ return ret;
ret = cz_dpm_force_lowest(adev);
if (ret)
return ret;
@@ -1711,6 +1755,8 @@ static int cz_dpm_force_dpm_level(struct amdgpu_device *adev,
break;
}
+ adev->pm.dpm.forced_level = level;
+
return ret;
}
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
index 5cde635978f9..e70a26f587a0 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c
@@ -2632,6 +2632,7 @@ static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
struct drm_device *dev = crtc->dev;
struct amdgpu_device *adev = dev->dev_private;
struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+ unsigned type;
switch (mode) {
case DRM_MODE_DPMS_ON:
@@ -2640,6 +2641,9 @@ static void dce_v10_0_crtc_dpms(struct drm_crtc *crtc, int mode)
dce_v10_0_vga_enable(crtc, true);
amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
dce_v10_0_vga_enable(crtc, false);
+ /* Make sure VBLANK interrupt is still enabled */
+ type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+ amdgpu_irq_update(adev, &adev->crtc_irq, type);
drm_vblank_post_modeset(dev, amdgpu_crtc->crtc_id);
dce_v10_0_crtc_load_lut(crtc);
break;
@@ -3403,19 +3407,25 @@ static int dce_v10_0_crtc_irq(struct amdgpu_device *adev,
switch (entry->src_data) {
case 0: /* vblank */
- if (disp_int & interrupt_status_offsets[crtc].vblank) {
+ if (disp_int & interrupt_status_offsets[crtc].vblank)
dce_v10_0_crtc_vblank_int_ack(adev, crtc);
- if (amdgpu_irq_enabled(adev, source, irq_type)) {
- drm_handle_vblank(adev->ddev, crtc);
- }
- DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+ else
+ DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+ if (amdgpu_irq_enabled(adev, source, irq_type)) {
+ drm_handle_vblank(adev->ddev, crtc);
}
+ DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
break;
case 1: /* vline */
- if (disp_int & interrupt_status_offsets[crtc].vline) {
+ if (disp_int & interrupt_status_offsets[crtc].vline)
dce_v10_0_crtc_vline_int_ack(adev, crtc);
- DRM_DEBUG("IH: D%d vline\n", crtc + 1);
- }
+ else
+ DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+ DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
break;
default:
DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
index 95efd98b202d..dcb402ee048a 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c
@@ -2631,6 +2631,7 @@ static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
struct drm_device *dev = crtc->dev;
struct amdgpu_device *adev = dev->dev_private;
struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+ unsigned type;
switch (mode) {
case DRM_MODE_DPMS_ON:
@@ -2639,6 +2640,9 @@ static void dce_v11_0_crtc_dpms(struct drm_crtc *crtc, int mode)
dce_v11_0_vga_enable(crtc, true);
amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
dce_v11_0_vga_enable(crtc, false);
+ /* Make sure VBLANK interrupt is still enabled */
+ type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+ amdgpu_irq_update(adev, &adev->crtc_irq, type);
drm_vblank_post_modeset(dev, amdgpu_crtc->crtc_id);
dce_v11_0_crtc_load_lut(crtc);
break;
@@ -3402,19 +3406,25 @@ static int dce_v11_0_crtc_irq(struct amdgpu_device *adev,
switch (entry->src_data) {
case 0: /* vblank */
- if (disp_int & interrupt_status_offsets[crtc].vblank) {
+ if (disp_int & interrupt_status_offsets[crtc].vblank)
dce_v11_0_crtc_vblank_int_ack(adev, crtc);
- if (amdgpu_irq_enabled(adev, source, irq_type)) {
- drm_handle_vblank(adev->ddev, crtc);
- }
- DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+ else
+ DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+ if (amdgpu_irq_enabled(adev, source, irq_type)) {
+ drm_handle_vblank(adev->ddev, crtc);
}
+ DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
break;
case 1: /* vline */
- if (disp_int & interrupt_status_offsets[crtc].vline) {
+ if (disp_int & interrupt_status_offsets[crtc].vline)
dce_v11_0_crtc_vline_int_ack(adev, crtc);
- DRM_DEBUG("IH: D%d vline\n", crtc + 1);
- }
+ else
+ DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+ DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
break;
default:
DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
index aaca8d663f2c..cc050a329c49 100644
--- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c
@@ -2566,6 +2566,7 @@ static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
struct drm_device *dev = crtc->dev;
struct amdgpu_device *adev = dev->dev_private;
struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
+ unsigned type;
switch (mode) {
case DRM_MODE_DPMS_ON:
@@ -2574,6 +2575,9 @@ static void dce_v8_0_crtc_dpms(struct drm_crtc *crtc, int mode)
dce_v8_0_vga_enable(crtc, true);
amdgpu_atombios_crtc_blank(crtc, ATOM_DISABLE);
dce_v8_0_vga_enable(crtc, false);
+ /* Make sure VBLANK interrupt is still enabled */
+ type = amdgpu_crtc_idx_to_irq_type(adev, amdgpu_crtc->crtc_id);
+ amdgpu_irq_update(adev, &adev->crtc_irq, type);
drm_vblank_post_modeset(dev, amdgpu_crtc->crtc_id);
dce_v8_0_crtc_load_lut(crtc);
break;
@@ -3237,19 +3241,25 @@ static int dce_v8_0_crtc_irq(struct amdgpu_device *adev,
switch (entry->src_data) {
case 0: /* vblank */
- if (disp_int & interrupt_status_offsets[crtc].vblank) {
+ if (disp_int & interrupt_status_offsets[crtc].vblank)
WREG32(mmLB_VBLANK_STATUS + crtc_offsets[crtc], LB_VBLANK_STATUS__VBLANK_ACK_MASK);
- if (amdgpu_irq_enabled(adev, source, irq_type)) {
- drm_handle_vblank(adev->ddev, crtc);
- }
- DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+ else
+ DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+ if (amdgpu_irq_enabled(adev, source, irq_type)) {
+ drm_handle_vblank(adev->ddev, crtc);
}
+ DRM_DEBUG("IH: D%d vblank\n", crtc + 1);
+
break;
case 1: /* vline */
- if (disp_int & interrupt_status_offsets[crtc].vline) {
+ if (disp_int & interrupt_status_offsets[crtc].vline)
WREG32(mmLB_VLINE_STATUS + crtc_offsets[crtc], LB_VLINE_STATUS__VLINE_ACK_MASK);
- DRM_DEBUG("IH: D%d vline\n", crtc + 1);
- }
+ else
+ DRM_DEBUG("IH: IH event w/o asserted irq bit?\n");
+
+ DRM_DEBUG("IH: D%d vline\n", crtc + 1);
+
break;
default:
DRM_DEBUG("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data);
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 7b683fb2173c..1c7c992dea37 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -1813,10 +1813,7 @@ static u32 gfx_v8_0_get_rb_disabled(struct amdgpu_device *adev,
u32 data, mask;
data = RREG32(mmCC_RB_BACKEND_DISABLE);
- if (data & 1)
- data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
- else
- data = 0;
+ data &= CC_RB_BACKEND_DISABLE__BACKEND_DISABLE_MASK;
data |= RREG32(mmGC_USER_RB_BACKEND_DISABLE);
diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
index d62c4002e39c..d1064ca3670e 100644
--- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
@@ -35,6 +35,8 @@
#include "oss/oss_2_0_d.h"
#include "oss/oss_2_0_sh_mask.h"
#include "gca/gfx_8_0_d.h"
+#include "smu/smu_7_1_2_d.h"
+#include "smu/smu_7_1_2_sh_mask.h"
#define GRBM_GFX_INDEX__VCE_INSTANCE__SHIFT 0x04
#define GRBM_GFX_INDEX__VCE_INSTANCE_MASK 0x10
@@ -112,6 +114,10 @@ static int vce_v3_0_start(struct amdgpu_device *adev)
mutex_lock(&adev->grbm_idx_mutex);
for (idx = 0; idx < 2; ++idx) {
+
+ if (adev->vce.harvest_config & (1 << idx))
+ continue;
+
if(idx == 0)
WREG32_P(mmGRBM_GFX_INDEX, 0,
~GRBM_GFX_INDEX__VCE_INSTANCE_MASK);
@@ -190,10 +196,52 @@ static int vce_v3_0_start(struct amdgpu_device *adev)
return 0;
}
+#define ixVCE_HARVEST_FUSE_MACRO__ADDRESS 0xC0014074
+#define VCE_HARVEST_FUSE_MACRO__SHIFT 27
+#define VCE_HARVEST_FUSE_MACRO__MASK 0x18000000
+
+static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev)
+{
+ u32 tmp;
+ unsigned ret;
+
+ if (adev->flags & AMDGPU_IS_APU)
+ tmp = (RREG32_SMC(ixVCE_HARVEST_FUSE_MACRO__ADDRESS) &
+ VCE_HARVEST_FUSE_MACRO__MASK) >>
+ VCE_HARVEST_FUSE_MACRO__SHIFT;
+ else
+ tmp = (RREG32_SMC(ixCC_HARVEST_FUSES) &
+ CC_HARVEST_FUSES__VCE_DISABLE_MASK) >>
+ CC_HARVEST_FUSES__VCE_DISABLE__SHIFT;
+
+ switch (tmp) {
+ case 1:
+ ret = AMDGPU_VCE_HARVEST_VCE0;
+ break;
+ case 2:
+ ret = AMDGPU_VCE_HARVEST_VCE1;
+ break;
+ case 3:
+ ret = AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1;
+ break;
+ default:
+ ret = 0;
+ }
+
+ return ret;
+}
+
static int vce_v3_0_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ adev->vce.harvest_config = vce_v3_0_get_harvest_config(adev);
+
+ if ((adev->vce.harvest_config &
+ (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1)) ==
+ (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1))
+ return -ENOENT;
+
vce_v3_0_set_ring_funcs(adev);
vce_v3_0_set_irq_funcs(adev);
diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index fa5a4448531d..68552da40287 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -122,6 +122,32 @@ static void vi_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
}
+/* smu_8_0_d.h */
+#define mmMP0PUB_IND_INDEX 0x180
+#define mmMP0PUB_IND_DATA 0x181
+
+static u32 cz_smc_rreg(struct amdgpu_device *adev, u32 reg)
+{
+ unsigned long flags;
+ u32 r;
+
+ spin_lock_irqsave(&adev->smc_idx_lock, flags);
+ WREG32(mmMP0PUB_IND_INDEX, (reg));
+ r = RREG32(mmMP0PUB_IND_DATA);
+ spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+ return r;
+}
+
+static void cz_smc_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&adev->smc_idx_lock, flags);
+ WREG32(mmMP0PUB_IND_INDEX, (reg));
+ WREG32(mmMP0PUB_IND_DATA, (v));
+ spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
+}
+
static u32 vi_uvd_ctx_rreg(struct amdgpu_device *adev, u32 reg)
{
unsigned long flags;
@@ -1222,8 +1248,13 @@ static int vi_common_early_init(void *handle)
bool smc_enabled = false;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- adev->smc_rreg = &vi_smc_rreg;
- adev->smc_wreg = &vi_smc_wreg;
+ if (adev->flags & AMDGPU_IS_APU) {
+ adev->smc_rreg = &cz_smc_rreg;
+ adev->smc_wreg = &cz_smc_wreg;
+ } else {
+ adev->smc_rreg = &vi_smc_rreg;
+ adev->smc_wreg = &vi_smc_wreg;
+ }
adev->pcie_rreg = &vi_pcie_rreg;
adev->pcie_wreg = &vi_pcie_wreg;
adev->uvd_ctx_rreg = &vi_uvd_ctx_rreg;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index 8a1f999daa24..9be007081b72 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -420,6 +420,12 @@ void kfd_unbind_process_from_device(struct kfd_dev *dev, unsigned int pasid)
pqm_uninit(&p->pqm);
pdd = kfd_get_process_device_data(dev, p);
+
+ if (!pdd) {
+ mutex_unlock(&p->mutex);
+ return;
+ }
+
if (pdd->reset_wavefronts) {
dbgdev_wave_reset_wavefronts(pdd->dev, p);
pdd->reset_wavefronts = false;
@@ -431,8 +437,7 @@ void kfd_unbind_process_from_device(struct kfd_dev *dev, unsigned int pasid)
* We don't call amd_iommu_unbind_pasid() here
* because the IOMMU called us.
*/
- if (pdd)
- pdd->bound = false;
+ pdd->bound = false;
mutex_unlock(&p->mutex);
}