From: Mukul Joshi mukul.joshi@amd.com
[ Upstream commit 97e3c6a853f2af9145daf0c6ca25bcdf55c759d4 ]
Currently, we store CU info only for a single XCC assuming that it is the same for all XCCs. However, that may not be true. As a result, store CU info for all XCCs. This info is later used for CU masking.
Signed-off-by: Mukul Joshi mukul.joshi@amd.com Reviewed-by: Felix Kuehling Felix.Kuehling@amd.com Signed-off-by: Alex Deucher alexander.deucher@amd.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h | 3 +- drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 4 +- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c | 76 +++++++++---------- drivers/gpu/drm/amd/amdkfd/kfd_crat.c | 3 +- drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c | 8 +- drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 11 ++- .../gpu/drm/amd/include/kgd_kfd_interface.h | 6 +- 14 files changed, 60 insertions(+), 65 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index b4fcad0e62f7e..a7c8beff1647c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -492,7 +492,7 @@ void amdgpu_amdkfd_get_cu_info(struct amdgpu_device *adev, struct kfd_cu_info *c cu_info->cu_active_number = acu_info.number; cu_info->cu_ao_mask = acu_info.ao_cu_mask; memcpy(&cu_info->cu_bitmap[0], &acu_info.bitmap[0], - sizeof(acu_info.bitmap)); + sizeof(cu_info->cu_bitmap)); cu_info->num_shader_engines = adev->gfx.config.max_shader_engines; cu_info->num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se; cu_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index a4ff515ce8966..59ba03d387fcc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -43,6 +43,7 @@ #define AMDGPU_GFX_LBPW_DISABLED_MODE 0x00000008L
#define AMDGPU_MAX_GC_INSTANCES 8 +#define KGD_MAX_QUEUES 128
#define AMDGPU_MAX_GFX_QUEUES KGD_MAX_QUEUES #define AMDGPU_MAX_COMPUTE_QUEUES KGD_MAX_QUEUES @@ -254,7 +255,7 @@ struct amdgpu_cu_info { uint32_t number; uint32_t ao_cu_mask; uint32_t ao_cu_bitmap[4][4]; - uint32_t bitmap[4][4]; + uint32_t bitmap[AMDGPU_MAX_GC_INSTANCES][4][4]; };
struct amdgpu_gfx_ras { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index d4ca19ba5a289..f678bdd5f353d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -839,7 +839,7 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) memcpy(&dev_info->cu_ao_bitmap[0], &adev->gfx.cu_info.ao_cu_bitmap[0], sizeof(adev->gfx.cu_info.ao_cu_bitmap)); memcpy(&dev_info->cu_bitmap[0], &adev->gfx.cu_info.bitmap[0], - sizeof(adev->gfx.cu_info.bitmap)); + sizeof(dev_info->cu_bitmap)); dev_info->vram_type = adev->gmc.vram_type; dev_info->vram_bit_width = adev->gmc.vram_width; dev_info->vce_harvest_config = adev->vce.harvest_config; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 44af8022b89fa..f743bf2c92877 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -9448,7 +9448,7 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev, gfx_v10_0_set_user_wgp_inactive_bitmap_per_sh( adev, disable_masks[i * 2 + j]); bitmap = gfx_v10_0_get_cu_active_bitmap_per_sh(adev); - cu_info->bitmap[i][j] = bitmap; + cu_info->bitmap[0][i][j] = bitmap;
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) { if (bitmap & mask) { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 0451533ddde41..a82cba884c48f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -6394,7 +6394,7 @@ static int gfx_v11_0_get_cu_info(struct amdgpu_device *adev, * SE6: {SH0,SH1} --> {bitmap[2][2], bitmap[2][3]} * SE7: {SH0,SH1} --> {bitmap[3][2], bitmap[3][3]} */ - cu_info->bitmap[i % 4][j + (i / 4) * 2] = bitmap; + cu_info->bitmap[0][i % 4][j + (i / 4) * 2] = bitmap;
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) { if (bitmap & mask) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index da6caff78c22b..34f9211b26793 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -3577,7 +3577,7 @@ static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev) gfx_v6_0_set_user_cu_inactive_bitmap( adev, disable_masks[i * 2 + j]); bitmap = gfx_v6_0_get_cu_enabled(adev); - cu_info->bitmap[i][j] = bitmap; + cu_info->bitmap[0][i][j] = bitmap;
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) { if (bitmap & mask) { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 8c174c11eaee0..6feae2548e8ee 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -5122,7 +5122,7 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev) gfx_v7_0_set_user_cu_inactive_bitmap( adev, disable_masks[i * 2 + j]); bitmap = gfx_v7_0_get_cu_active_bitmap(adev); - cu_info->bitmap[i][j] = bitmap; + cu_info->bitmap[0][i][j] = bitmap;
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) { if (bitmap & mask) { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 51c1745c83697..885ebd703260f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -7121,7 +7121,7 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev) gfx_v8_0_set_user_cu_inactive_bitmap( adev, disable_masks[i * 2 + j]); bitmap = gfx_v8_0_get_cu_active_bitmap(adev); - cu_info->bitmap[i][j] = bitmap; + cu_info->bitmap[0][i][j] = bitmap;
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) { if (bitmap & mask) { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 65577eca58f1c..e511e49d3023f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -1497,7 +1497,7 @@ static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev) amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) { - if (cu_info->bitmap[i][j] & mask) { + if (cu_info->bitmap[0][i][j] & mask) { if (counter == pg_always_on_cu_num) WREG32_SOC15(GC, 0, mmRLC_PG_ALWAYS_ON_CU_MASK, cu_bitmap); if (counter < always_on_cu_num) @@ -7234,7 +7234,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, * SE6,SH0 --> bitmap[2][1] * SE7,SH0 --> bitmap[3][1] */ - cu_info->bitmap[i % 4][j + i / 4] = bitmap; + cu_info->bitmap[0][i % 4][j + i / 4] = bitmap;
for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) { if (bitmap & mask) { diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index 4f883b94f98ef..84a74a6c6b2de 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -4228,7 +4228,7 @@ static void gfx_v9_4_3_set_gds_init(struct amdgpu_device *adev) }
static void gfx_v9_4_3_set_user_cu_inactive_bitmap(struct amdgpu_device *adev, - u32 bitmap) + u32 bitmap, int xcc_id) { u32 data;
@@ -4238,15 +4238,15 @@ static void gfx_v9_4_3_set_user_cu_inactive_bitmap(struct amdgpu_device *adev, data = bitmap << GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT; data &= GC_USER_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK;
- WREG32_SOC15(GC, GET_INST(GC, 0), regGC_USER_SHADER_ARRAY_CONFIG, data); + WREG32_SOC15(GC, GET_INST(GC, xcc_id), regGC_USER_SHADER_ARRAY_CONFIG, data); }
-static u32 gfx_v9_4_3_get_cu_active_bitmap(struct amdgpu_device *adev) +static u32 gfx_v9_4_3_get_cu_active_bitmap(struct amdgpu_device *adev, int xcc_id) { u32 data, mask;
- data = RREG32_SOC15(GC, GET_INST(GC, 0), regCC_GC_SHADER_ARRAY_CONFIG); - data |= RREG32_SOC15(GC, GET_INST(GC, 0), regGC_USER_SHADER_ARRAY_CONFIG); + data = RREG32_SOC15(GC, GET_INST(GC, xcc_id), regCC_GC_SHADER_ARRAY_CONFIG); + data |= RREG32_SOC15(GC, GET_INST(GC, xcc_id), regGC_USER_SHADER_ARRAY_CONFIG);
data &= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS_MASK; data >>= CC_GC_SHADER_ARRAY_CONFIG__INACTIVE_CUS__SHIFT; @@ -4259,7 +4259,7 @@ static u32 gfx_v9_4_3_get_cu_active_bitmap(struct amdgpu_device *adev) static int gfx_v9_4_3_get_cu_info(struct amdgpu_device *adev, struct amdgpu_cu_info *cu_info) { - int i, j, k, counter, active_cu_number = 0; + int i, j, k, counter, xcc_id, active_cu_number = 0; u32 mask, bitmap, ao_bitmap, ao_cu_mask = 0; unsigned disable_masks[4 * 4];
@@ -4278,46 +4278,38 @@ static int gfx_v9_4_3_get_cu_info(struct amdgpu_device *adev, adev->gfx.config.max_sh_per_se);
mutex_lock(&adev->grbm_idx_mutex); - for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { - for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { - mask = 1; - ao_bitmap = 0; - counter = 0; - gfx_v9_4_3_xcc_select_se_sh(adev, i, j, 0xffffffff, 0); - gfx_v9_4_3_set_user_cu_inactive_bitmap( - adev, disable_masks[i * adev->gfx.config.max_sh_per_se + j]); - bitmap = gfx_v9_4_3_get_cu_active_bitmap(adev); - - /* - * The bitmap(and ao_cu_bitmap) in cu_info structure is - * 4x4 size array, and it's usually suitable for Vega - * ASICs which has 4*2 SE/SH layout. - * But for Arcturus, SE/SH layout is changed to 8*1. - * To mostly reduce the impact, we make it compatible - * with current bitmap array as below: - * SE4,SH0 --> bitmap[0][1] - * SE5,SH0 --> bitmap[1][1] - * SE6,SH0 --> bitmap[2][1] - * SE7,SH0 --> bitmap[3][1] - */ - cu_info->bitmap[i % 4][j + i / 4] = bitmap; - - for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) { - if (bitmap & mask) { - if (counter < adev->gfx.config.max_cu_per_sh) - ao_bitmap |= mask; - counter++; + for (xcc_id = 0; xcc_id < NUM_XCC(adev->gfx.xcc_mask); xcc_id++) { + for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { + for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { + mask = 1; + ao_bitmap = 0; + counter = 0; + gfx_v9_4_3_xcc_select_se_sh(adev, i, j, 0xffffffff, xcc_id); + gfx_v9_4_3_set_user_cu_inactive_bitmap( + adev, + disable_masks[i * adev->gfx.config.max_sh_per_se + j], + xcc_id); + bitmap = gfx_v9_4_3_get_cu_active_bitmap(adev, xcc_id); + + cu_info->bitmap[xcc_id][i][j] = bitmap; + + for (k = 0; k < adev->gfx.config.max_cu_per_sh; k++) { + if (bitmap & mask) { + if (counter < adev->gfx.config.max_cu_per_sh) + ao_bitmap |= mask; + counter++; + } + mask <<= 1; } - mask <<= 1; + active_cu_number += counter; + if (i < 2 && j < 2) + ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8)); + cu_info->ao_cu_bitmap[i][j] = ao_bitmap; } - active_cu_number += counter; - if (i < 2 && j < 2) - ao_cu_mask |= (ao_bitmap << (i * 16 + j * 8)); - cu_info->ao_cu_bitmap[i % 4][j + i / 4] = ao_bitmap; } + gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, + xcc_id); } - gfx_v9_4_3_xcc_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, - 0); mutex_unlock(&adev->grbm_idx_mutex);
cu_info->number = active_cu_number; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c index f5a6f562e2a80..11b9837292536 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c @@ -2154,7 +2154,8 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image,
amdgpu_amdkfd_get_cu_info(kdev->adev, &cu_info); cu->num_simd_per_cu = cu_info.simd_per_cu; - cu->num_simd_cores = cu_info.simd_per_cu * cu_info.cu_active_number; + cu->num_simd_cores = cu_info.simd_per_cu * + (cu_info.cu_active_number / kdev->kfd->num_nodes); cu->max_waves_simd = cu_info.max_waves_per_simd;
cu->wave_front_size = cu_info.wave_front_size; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c index 863cf060af484..35e05ee89eac5 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c @@ -104,11 +104,13 @@ void mqd_symmetrically_map_cu_mask(struct mqd_manager *mm, bool wgp_mode_req = KFD_GC_VERSION(mm->dev) >= IP_VERSION(10, 0, 0); uint32_t en_mask = wgp_mode_req ? 0x3 : 0x1; int i, se, sh, cu, cu_bitmap_sh_mul, inc = wgp_mode_req ? 2 : 1; + uint32_t cu_active_per_node;
amdgpu_amdkfd_get_cu_info(mm->dev->adev, &cu_info);
- if (cu_mask_count > cu_info.cu_active_number) - cu_mask_count = cu_info.cu_active_number; + cu_active_per_node = cu_info.cu_active_number / mm->dev->kfd->num_nodes; + if (cu_mask_count > cu_active_per_node) + cu_mask_count = cu_active_per_node;
/* Exceeding these bounds corrupts the stack and indicates a coding error. * Returning with no CU's enabled will hang the queue, which should be @@ -141,7 +143,7 @@ void mqd_symmetrically_map_cu_mask(struct mqd_manager *mm, for (se = 0; se < cu_info.num_shader_engines; se++) for (sh = 0; sh < cu_info.num_shader_arrays_per_engine; sh++) cu_per_sh[se][sh] = hweight32( - cu_info.cu_bitmap[se % 4][sh + (se / 4) * cu_bitmap_sh_mul]); + cu_info.cu_bitmap[0][se % 4][sh + (se / 4) * cu_bitmap_sh_mul]);
/* Symmetrically map cu_mask to all SEs & SHs: * se_mask programs up to 2 SH in the upper and lower 16 bits. diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index 4a17bb7c7b27d..ea67a353beb00 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -450,8 +450,7 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr, sysfs_show_32bit_prop(buffer, offs, "cpu_cores_count", dev->node_props.cpu_cores_count); sysfs_show_32bit_prop(buffer, offs, "simd_count", - dev->gpu ? (dev->node_props.simd_count * - NUM_XCC(dev->gpu->xcc_mask)) : 0); + dev->gpu ? dev->node_props.simd_count : 0); sysfs_show_32bit_prop(buffer, offs, "mem_banks_count", dev->node_props.mem_banks_count); sysfs_show_32bit_prop(buffer, offs, "caches_count", @@ -1658,7 +1657,7 @@ static int fill_in_l2_l3_pcache(struct kfd_cache_properties **props_ext, int i, j, k; struct kfd_cache_properties *pcache = NULL;
- cu_sibling_map_mask = cu_info->cu_bitmap[0][0]; + cu_sibling_map_mask = cu_info->cu_bitmap[0][0][0]; cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1); first_active_cu = ffs(cu_sibling_map_mask); @@ -1701,7 +1700,7 @@ static int fill_in_l2_l3_pcache(struct kfd_cache_properties **props_ext, pcache->sibling_map[k+3] = (uint8_t)((cu_sibling_map_mask >> 24) & 0xFF); k += 4;
- cu_sibling_map_mask = cu_info->cu_bitmap[i % 4][j + i / 4]; + cu_sibling_map_mask = cu_info->cu_bitmap[0][i % 4][j + i / 4]; cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1); } } @@ -1762,8 +1761,8 @@ static void kfd_fill_cache_non_crat_info(struct kfd_topology_device *dev, struct for (k = 0; k < pcu_info->num_cu_per_sh; k += pcache_info[ct].num_cu_shared) {
ret = fill_in_l1_pcache(&props_ext, pcache_info, pcu_info, - pcu_info->cu_bitmap[i % 4][j + i / 4], ct, - cu_processor_id, k); + pcu_info->cu_bitmap[0][i % 4][j + i / 4], ct, + cu_processor_id, k);
if (ret < 0) break; diff --git a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h index d0df3381539f0..74cc545085a02 100644 --- a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h @@ -31,12 +31,12 @@ #include <linux/types.h> #include <linux/bitmap.h> #include <linux/dma-fence.h> +#include "amdgpu_irq.h" +#include "amdgpu_gfx.h"
struct pci_dev; struct amdgpu_device;
-#define KGD_MAX_QUEUES 128 - struct kfd_dev; struct kgd_mem;
@@ -68,7 +68,7 @@ struct kfd_cu_info { uint32_t wave_front_size; uint32_t max_scratch_slots_per_cu; uint32_t lds_size; - uint32_t cu_bitmap[4][4]; + uint32_t cu_bitmap[AMDGPU_MAX_GC_INSTANCES][4][4]; };
/* For getting GPU local memory information from KGD */