drm/amdgpu: add xcc index argument to select_sh_se function v2

v1: To support multiple XCD case (Le)
v2: introduce xcc index to gfx_v11_0_select_sh_se (Hawking)

Signed-off-by: Le Ma <le.ma@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
index ebb35633058c24a7219a42d28181429d25c67ec4..ae06d1f2af93203d33ee04f4d5d6e7ae43657d04 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -787,7 +787,7 @@ void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev, int pasid,
        for (se_idx = 0; se_idx < se_cnt; se_idx++) {
                for (sh_idx = 0; sh_idx < sh_cnt; sh_idx++) {
 
-                       amdgpu_gfx_select_se_sh(adev, se_idx, sh_idx, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, se_idx, sh_idx, 0xffffffff, 0);
                        queue_map = RREG32_SOC15(GC, 0, mmSPI_CSQ_WF_ACTIVE_STATUS);
 
                        /*
@@ -820,7 +820,7 @@ void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev, int pasid,
                }
        }
 
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        soc15_grbm_select(adev, 0, 0, 0, 0, 0);
        unlock_spi_csq_mutexes(adev);
 

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
index f60753f97ac598bc04b617ba1e9ab7193cc58b64..cc64ae550bc1e0b6f45f2854b475bad5b558eb5b 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
@@ -136,7 +136,7 @@ static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
                }
                mutex_lock(&adev->grbm_idx_mutex);
                amdgpu_gfx_select_se_sh(adev, se_bank,
-                                       sh_bank, instance_bank);
+                                       sh_bank, instance_bank, 0);
        } else if (use_ring) {
                mutex_lock(&adev->srbm_mutex);
                amdgpu_gfx_select_me_pipe_q(adev, me, pipe, queue, vmid);
@@ -169,7 +169,7 @@ static int  amdgpu_debugfs_process_reg_op(bool read, struct file *f,
 
 end:
        if (use_bank) {
-               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                mutex_unlock(&adev->grbm_idx_mutex);
        } else if (use_ring) {
                amdgpu_gfx_select_me_pipe_q(adev, 0, 0, 0, 0);
@@ -263,7 +263,7 @@ static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 off
                mutex_lock(&adev->grbm_idx_mutex);
                amdgpu_gfx_select_se_sh(adev, rd->id.grbm.se,
                                                                rd->id.grbm.sh,
-                                                               rd->id.grbm.instance);
+                                                               rd->id.grbm.instance, 0);
        }
 
        if (rd->id.use_srbm) {
@@ -295,7 +295,7 @@ static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 off
        }
 end:
        if (rd->id.use_grbm) {
-               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                mutex_unlock(&adev->grbm_idx_mutex);
        }
 
@@ -907,13 +907,13 @@ static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf,
 
        /* switch to the specific se/sh/cu */
        mutex_lock(&adev->grbm_idx_mutex);
-       amdgpu_gfx_select_se_sh(adev, se, sh, cu);
+       amdgpu_gfx_select_se_sh(adev, se, sh, cu, 0);
 
        x = 0;
        if (adev->gfx.funcs->read_wave_data)
                adev->gfx.funcs->read_wave_data(adev, simd, wave, data, &x);
 
-       amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+       amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);
@@ -1001,7 +1001,7 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
 
        /* switch to the specific se/sh/cu */
        mutex_lock(&adev->grbm_idx_mutex);
-       amdgpu_gfx_select_se_sh(adev, se, sh, cu);
+       amdgpu_gfx_select_se_sh(adev, se, sh, cu, 0);
 
        if (bank == 0) {
                if (adev->gfx.funcs->read_wave_vgprs)
@@ -1011,7 +1011,7 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf,
                        adev->gfx.funcs->read_wave_sgprs(adev, simd, wave, offset, size>>2, data);
        }
 
-       amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+       amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        pm_runtime_mark_last_busy(adev_to_drm(adev)->dev);

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index a9e41d7970eadff1f6b704897464f13835be42d3..77e2f714e357daaff50cc37bda23d91cd8554fab 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -237,7 +237,7 @@ struct amdgpu_gfx_funcs {
        /* get the gpu clock counter */
        uint64_t (*get_gpu_clock_counter)(struct amdgpu_device *adev);
        void (*select_se_sh)(struct amdgpu_device *adev, u32 se_num,
-                            u32 sh_num, u32 instance);
+                            u32 sh_num, u32 instance, int xcc_id);
        void (*read_wave_data)(struct amdgpu_device *adev, uint32_t simd,
                               uint32_t wave, uint32_t *dst, int *no_fields);
        void (*read_wave_vgprs)(struct amdgpu_device *adev, uint32_t simd,
@@ -386,7 +386,7 @@ struct amdgpu_gfx {
 };
 
 #define amdgpu_gfx_get_gpu_clock_counter(adev) (adev)->gfx.funcs->get_gpu_clock_counter((adev))
-#define amdgpu_gfx_select_se_sh(adev, se, sh, instance) (adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance))
+#define amdgpu_gfx_select_se_sh(adev, se, sh, instance, xcc_id) ((adev)->gfx.funcs->select_se_sh((adev), (se), (sh), (instance), (xcc_id)))
 #define amdgpu_gfx_select_me_pipe_q(adev, me, pipe, q, vmid) (adev)->gfx.funcs->select_me_pipe_q((adev), (me), (pipe), (q), (vmid))
 #define amdgpu_gfx_init_spm_golden(adev) (adev)->gfx.funcs->init_spm_golden((adev))
 

diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index de6d10390ab2f195fcb3a03091f904ce1edf2c0c..5641cf05d856b4253b9086376b720da92ad4ba64 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -1141,12 +1141,12 @@ static uint32_t cik_get_register_value(struct amdgpu_device *adev,
 
                mutex_lock(&adev->grbm_idx_mutex);
                if (se_num != 0xffffffff || sh_num != 0xffffffff)
-                       amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
                val = RREG32(reg_offset);
 
                if (se_num != 0xffffffff || sh_num != 0xffffffff)
-                       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                mutex_unlock(&adev->grbm_idx_mutex);
                return val;
        } else {

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index accc0a7251b9aaff25b564219c66d90aa264da81..323f5b8927ad114edc77beaee78e4a05f0fb44eb 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -3490,7 +3490,7 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
                                 struct amdgpu_cu_info *cu_info);
 static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev);
 static void gfx_v10_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
-                                  u32 sh_num, u32 instance);
+                                  u32 sh_num, u32 instance, int xcc_id);
 static u32 gfx_v10_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev);
 
 static int gfx_v10_0_rlc_backdoor_autoload_buffer_init(struct amdgpu_device *adev);
@@ -4712,7 +4712,7 @@ static int gfx_v10_0_sw_fini(void *handle)
 }
 
 static void gfx_v10_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
-                                  u32 sh_num, u32 instance)
+                                  u32 sh_num, u32 instance, int xcc_id)
 {
        u32 data;
 
@@ -4772,13 +4772,13 @@ static void gfx_v10_0_setup_rb(struct amdgpu_device *adev)
                                (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 6))) &&
                            ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1))
                                continue;
-                       gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        data = gfx_v10_0_get_rb_active_bitmap(adev);
                        active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
                                               rb_bitmap_width_per_sh);
                }
        }
-       gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        adev->gfx.config.backend_enable_mask = active_rbs;
@@ -4907,7 +4907,7 @@ static void gfx_v10_0_tcp_harvest(struct amdgpu_device *adev)
        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++) {
-                       gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        wgp_active_bitmap = gfx_v10_0_get_wgp_active_bitmap_per_sh(adev);
                        /*
                         * Set corresponding TCP bits for the inactive WGPs in
@@ -4940,7 +4940,7 @@ static void gfx_v10_0_tcp_harvest(struct amdgpu_device *adev)
                }
        }
 
-       gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -9540,7 +9540,7 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
                        mask = 1;
                        ao_bitmap = 0;
                        counter = 0;
-                       gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        if (i < 4 && j < 2)
                                gfx_v10_0_set_user_wgp_inactive_bitmap_per_sh(
                                        adev, disable_masks[i * 2 + j]);
@@ -9561,7 +9561,7 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
                        cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
                }
        }
-       gfx_v10_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v10_0_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/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index 4fbefe236fc715b40b994e2670fb1067933ed301..50d0ff9ca25993b9a463fbe7fbf13856cdeb736e 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -112,7 +112,7 @@ static int gfx_v11_0_get_cu_info(struct amdgpu_device *adev,
                                  struct amdgpu_cu_info *cu_info);
 static uint64_t gfx_v11_0_get_gpu_clock_counter(struct amdgpu_device *adev);
 static void gfx_v11_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
-                                  u32 sh_num, u32 instance);
+                                  u32 sh_num, u32 instance, int xcc_id);
 static u32 gfx_v11_0_get_wgp_active_bitmap_per_sh(struct amdgpu_device *adev);
 
 static void gfx_v11_0_ring_emit_de_meta(struct amdgpu_ring *ring, bool resume);
@@ -1484,7 +1484,7 @@ static int gfx_v11_0_sw_fini(void *handle)
 }
 
 static void gfx_v11_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
-                                  u32 sh_num, u32 instance)
+                                  u32 sh_num, u32 instance, int xcc_id)
 {
        u32 data;
 
@@ -6473,7 +6473,7 @@ static int gfx_v11_0_get_cu_info(struct amdgpu_device *adev,
                for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
                        mask = 1;
                        counter = 0;
-                       gfx_v11_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v11_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        if (i < 8 && j < 2)
                                gfx_v11_0_set_user_wgp_inactive_bitmap_per_sh(
                                        adev, disable_masks[i * 2 + j]);
@@ -6505,7 +6505,7 @@ static int gfx_v11_0_get_cu_info(struct amdgpu_device *adev,
                        active_cu_number += counter;
                }
        }
-       gfx_v11_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v11_0_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/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
index d9ce4d1c50e496388d7c585625667dbec42d6207..7cb72bf1acdd393b5d017c4573eb387dbae580e7 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
@@ -1285,7 +1285,7 @@ static void gfx_v6_0_tiling_mode_table_init(struct amdgpu_device *adev)
 }
 
 static void gfx_v6_0_select_se_sh(struct amdgpu_device *adev, u32 se_num,
-                                 u32 sh_num, u32 instance)
+                                 u32 sh_num, u32 instance, int xcc_id)
 {
        u32 data;
 
@@ -1438,12 +1438,12 @@ static void gfx_v6_0_write_harvested_raster_configs(struct amdgpu_device *adev,
                }
 
                /* GRBM_GFX_INDEX has a different offset on SI */
-               gfx_v6_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+               gfx_v6_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff, 0);
                WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
        }
 
        /* GRBM_GFX_INDEX has a different offset on SI */
-       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 }
 
 static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
@@ -1459,14 +1459,14 @@ static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
        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++) {
-                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        data = gfx_v6_0_get_rb_active_bitmap(adev);
                        active_rbs |= data <<
                                ((i * adev->gfx.config.max_sh_per_se + j) *
                                 rb_bitmap_width_per_sh);
                }
        }
-       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        adev->gfx.config.backend_enable_mask = active_rbs;
        adev->gfx.config.num_rbs = hweight32(active_rbs);
@@ -1487,7 +1487,7 @@ static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
        /* cache the values for userspace */
        for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
                for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        adev->gfx.config.rb_config[i][j].rb_backend_disable =
                                RREG32(mmCC_RB_BACKEND_DISABLE);
                        adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
@@ -1496,7 +1496,7 @@ static void gfx_v6_0_setup_rb(struct amdgpu_device *adev)
                                RREG32(mmPA_SC_RASTER_CONFIG);
                }
        }
-       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -1535,7 +1535,7 @@ static void gfx_v6_0_setup_spi(struct amdgpu_device *adev)
        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++) {
-                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        data = RREG32(mmSPI_STATIC_THREAD_MGMT_3);
                        active_cu = gfx_v6_0_get_cu_enabled(adev);
 
@@ -1550,7 +1550,7 @@ static void gfx_v6_0_setup_spi(struct amdgpu_device *adev)
                        }
                }
        }
-       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -2391,7 +2391,7 @@ static void gfx_v6_0_enable_lbpw(struct amdgpu_device *adev, bool enable)
        WREG32_FIELD(RLC_LB_CNTL, LOAD_BALANCE_ENABLE, enable ? 1 : 0);
 
        if (!enable) {
-               gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                WREG32(mmSPI_LB_CU_MASK, 0x00ff);
        }
 }
@@ -3571,7 +3571,7 @@ static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev)
                        mask = 1;
                        ao_bitmap = 0;
                        counter = 0;
-                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v6_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        if (i < 4 && j < 2)
                                gfx_v6_0_set_user_cu_inactive_bitmap(
                                        adev, disable_masks[i * 2 + j]);
@@ -3593,7 +3593,7 @@ static void gfx_v6_0_get_cu_info(struct amdgpu_device *adev)
                }
        }
 
-       gfx_v6_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v6_0_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/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 46740ad9a80ff4d5b4c923df0e9fd32e1744faf6..d055e44eee1d80bc58cd514e8ce785711cf49496 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -1552,7 +1552,8 @@ static void gfx_v7_0_tiling_mode_table_init(struct amdgpu_device *adev)
  * Select which SE, SH combinations to address.
  */
 static void gfx_v7_0_select_se_sh(struct amdgpu_device *adev,
-                                 u32 se_num, u32 sh_num, u32 instance)
+                                 u32 se_num, u32 sh_num, u32 instance,
+                                 int xcc_id)
 {
        u32 data;
 
@@ -1732,13 +1733,13 @@ gfx_v7_0_write_harvested_raster_configs(struct amdgpu_device *adev,
                }
 
                /* GRBM_GFX_INDEX has a different offset on CI+ */
-               gfx_v7_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+               gfx_v7_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff, 0);
                WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
                WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
        }
 
        /* GRBM_GFX_INDEX has a different offset on CI+ */
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 }
 
 /**
@@ -1761,13 +1762,13 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
        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++) {
-                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        data = gfx_v7_0_get_rb_active_bitmap(adev);
                        active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
                                               rb_bitmap_width_per_sh);
                }
        }
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        adev->gfx.config.backend_enable_mask = active_rbs;
        adev->gfx.config.num_rbs = hweight32(active_rbs);
@@ -1790,7 +1791,7 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
        /* cache the values for userspace */
        for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
                for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        adev->gfx.config.rb_config[i][j].rb_backend_disable =
                                RREG32(mmCC_RB_BACKEND_DISABLE);
                        adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
@@ -1801,7 +1802,7 @@ static void gfx_v7_0_setup_rb(struct amdgpu_device *adev)
                                RREG32(mmPA_SC_RASTER_CONFIG_1);
                }
        }
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -1911,7 +1912,7 @@ static void gfx_v7_0_constants_init(struct amdgpu_device *adev)
         * making sure that the following register writes will be broadcasted
         * to all the shaders
         */
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        /* XXX SH_MEM regs */
        /* where to put LDS, scratch, GPUVM in FSA64 space */
@@ -3301,7 +3302,7 @@ static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
        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++) {
-                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        for (k = 0; k < adev->usec_timeout; k++) {
                                if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
                                        break;
@@ -3309,7 +3310,7 @@ static void gfx_v7_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
                        }
                }
        }
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -3474,7 +3475,7 @@ static int gfx_v7_0_rlc_resume(struct amdgpu_device *adev)
        WREG32(mmRLC_LB_CNTR_MAX, 0x00008000);
 
        mutex_lock(&adev->grbm_idx_mutex);
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        WREG32(mmRLC_LB_INIT_CU_MASK, 0xffffffff);
        WREG32(mmRLC_LB_PARAMS, 0x00600408);
        WREG32(mmRLC_LB_CNTL, 0x80000004);
@@ -3530,7 +3531,7 @@ static void gfx_v7_0_enable_cgcg(struct amdgpu_device *adev, bool enable)
                tmp = gfx_v7_0_halt_rlc(adev);
 
                mutex_lock(&adev->grbm_idx_mutex);
-               gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
                WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
                tmp2 = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
@@ -3584,7 +3585,7 @@ static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
                tmp = gfx_v7_0_halt_rlc(adev);
 
                mutex_lock(&adev->grbm_idx_mutex);
-               gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
                WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
                data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK |
@@ -3635,7 +3636,7 @@ static void gfx_v7_0_enable_mgcg(struct amdgpu_device *adev, bool enable)
                tmp = gfx_v7_0_halt_rlc(adev);
 
                mutex_lock(&adev->grbm_idx_mutex);
-               gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
                WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
                data = RLC_SERDES_WR_CTRL__BPM_ADDR_MASK | RLC_SERDES_WR_CTRL__MGCG_OVERRIDE_1_MASK;
@@ -5115,7 +5116,7 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
                        mask = 1;
                        ao_bitmap = 0;
                        counter = 0;
-                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v7_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        if (i < 4 && j < 2)
                                gfx_v7_0_set_user_cu_inactive_bitmap(
                                        adev, disable_masks[i * 2 + j]);
@@ -5136,7 +5137,7 @@ static void gfx_v7_0_get_cu_info(struct amdgpu_device *adev)
                        cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
                }
        }
-       gfx_v7_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v7_0_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/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 8a43e87de49fa412ec4d8304b67bd8db8f37ef59..b604808761499dde387d20d4b05fb2d7c8275034 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -3395,7 +3395,8 @@ static void gfx_v8_0_tiling_mode_table_init(struct amdgpu_device *adev)
 }
 
 static void gfx_v8_0_select_se_sh(struct amdgpu_device *adev,
-                                 u32 se_num, u32 sh_num, u32 instance)
+                                 u32 se_num, u32 sh_num, u32 instance,
+                                 int xcc_id)
 {
        u32 data;
 
@@ -3579,13 +3580,13 @@ gfx_v8_0_write_harvested_raster_configs(struct amdgpu_device *adev,
                }
 
                /* GRBM_GFX_INDEX has a different offset on VI */
-               gfx_v8_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff);
+               gfx_v8_0_select_se_sh(adev, se, 0xffffffff, 0xffffffff, 0);
                WREG32(mmPA_SC_RASTER_CONFIG, raster_config_se);
                WREG32(mmPA_SC_RASTER_CONFIG_1, raster_config_1);
        }
 
        /* GRBM_GFX_INDEX has a different offset on VI */
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 }
 
 static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
@@ -3601,13 +3602,13 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
        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++) {
-                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        data = gfx_v8_0_get_rb_active_bitmap(adev);
                        active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
                                               rb_bitmap_width_per_sh);
                }
        }
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        adev->gfx.config.backend_enable_mask = active_rbs;
        adev->gfx.config.num_rbs = hweight32(active_rbs);
@@ -3630,7 +3631,7 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
        /* cache the values for userspace */
        for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
                for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
-                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        adev->gfx.config.rb_config[i][j].rb_backend_disable =
                                RREG32(mmCC_RB_BACKEND_DISABLE);
                        adev->gfx.config.rb_config[i][j].user_rb_backend_disable =
@@ -3641,7 +3642,7 @@ static void gfx_v8_0_setup_rb(struct amdgpu_device *adev)
                                RREG32(mmPA_SC_RASTER_CONFIG_1);
                }
        }
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -3788,7 +3789,7 @@ static void gfx_v8_0_constants_init(struct amdgpu_device *adev)
         * making sure that the following register writes will be broadcasted
         * to all the shaders
         */
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        WREG32(mmPA_SC_FIFO_SIZE,
                   (adev->gfx.config.sc_prim_fifo_size_frontend <<
@@ -3819,7 +3820,7 @@ static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
        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++) {
-                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        for (k = 0; k < adev->usec_timeout; k++) {
                                if (RREG32(mmRLC_SERDES_CU_MASTER_BUSY) == 0)
                                        break;
@@ -3827,7 +3828,7 @@ static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
                        }
                        if (k == adev->usec_timeout) {
                                gfx_v8_0_select_se_sh(adev, 0xffffffff,
-                                                     0xffffffff, 0xffffffff);
+                                                     0xffffffff, 0xffffffff, 0);
                                mutex_unlock(&adev->grbm_idx_mutex);
                                DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
                                         i, j);
@@ -3835,7 +3836,7 @@ static void gfx_v8_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
                        }
                }
        }
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -5481,7 +5482,7 @@ static void gfx_v8_0_send_serdes_cmd(struct amdgpu_device *adev,
 {
        uint32_t data;
 
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        WREG32(mmRLC_SERDES_WR_CU_MASTER_MASK, 0xffffffff);
        WREG32(mmRLC_SERDES_WR_NONCU_MASTER_MASK, 0xffffffff);
@@ -6723,11 +6724,11 @@ static void gfx_v8_0_parse_sq_irq(struct amdgpu_device *adev, unsigned ih_data,
                         */
                        if (from_wq) {
                                mutex_lock(&adev->grbm_idx_mutex);
-                               gfx_v8_0_select_se_sh(adev, se_id, sh_id, cu_id);
+                               gfx_v8_0_select_se_sh(adev, se_id, sh_id, cu_id, 0);
 
                                sq_edc_source = REG_GET_FIELD(RREG32(mmSQ_EDC_INFO), SQ_EDC_INFO, SOURCE);
 
-                               gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+                               gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                                mutex_unlock(&adev->grbm_idx_mutex);
                        }
 
@@ -7116,7 +7117,7 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
                        mask = 1;
                        ao_bitmap = 0;
                        counter = 0;
-                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v8_0_select_se_sh(adev, i, j, 0xffffffff, 0);
                        if (i < 4 && j < 2)
                                gfx_v8_0_set_user_cu_inactive_bitmap(
                                        adev, disable_masks[i * 2 + j]);
@@ -7137,7 +7138,7 @@ static void gfx_v8_0_get_cu_info(struct amdgpu_device *adev)
                        cu_info->ao_cu_bitmap[i][j] = ao_bitmap;
                }
        }
-       gfx_v8_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v8_0_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/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index 4939fd61355be732097fe669f11c3912b345bbdd..2fa7adef18a9100cd589fefe3835efa1b521d595 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -1504,7 +1504,7 @@ static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev)
                        mask = 1;
                        cu_bitmap = 0;
                        counter = 0;
-                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff);
+                       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) {
@@ -1523,7 +1523,7 @@ static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev)
                        cu_info->ao_cu_bitmap[i][j] = cu_bitmap;
                }
        }
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 }
 
@@ -1545,7 +1545,7 @@ static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev)
 
        mutex_lock(&adev->grbm_idx_mutex);
        /* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff);
 
        /* set mmRLC_LB_PARAMS = 0x003F_1006 */
@@ -1594,7 +1594,7 @@ static void gfx_v9_4_init_lbpw(struct amdgpu_device *adev)
 
        mutex_lock(&adev->grbm_idx_mutex);
        /* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff);
 
        /* set mmRLC_LB_PARAMS = 0x003F_1006 */
@@ -2241,7 +2241,7 @@ static void gfx_v9_0_tiling_mode_table_init(struct amdgpu_device *adev)
 }
 
 void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num,
-                          u32 instance)
+                          u32 instance, int xcc_id)
 {
        u32 data;
 
@@ -2290,13 +2290,13 @@ static void gfx_v9_0_setup_rb(struct amdgpu_device *adev)
        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++) {
-                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
                        data = gfx_v9_0_get_rb_active_bitmap(adev);
                        active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
                                               rb_bitmap_width_per_sh);
                }
        }
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        adev->gfx.config.backend_enable_mask = active_rbs;
@@ -2433,7 +2433,7 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
        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++) {
-                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
                        for (k = 0; k < adev->usec_timeout; k++) {
                                if (RREG32_SOC15(GC, 0, mmRLC_SERDES_CU_MASTER_BUSY) == 0)
                                        break;
@@ -2441,7 +2441,7 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
                        }
                        if (k == adev->usec_timeout) {
                                amdgpu_gfx_select_se_sh(adev, 0xffffffff,
-                                                     0xffffffff, 0xffffffff);
+                                                     0xffffffff, 0xffffffff, 0);
                                mutex_unlock(&adev->grbm_idx_mutex);
                                DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
                                         i, j);
@@ -2449,7 +2449,7 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev)
                        }
                }
        }
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -6608,7 +6608,7 @@ static void gfx_v9_0_reset_ras_error_count(struct amdgpu_device *adev)
        for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {
                for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) {
                        for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) {
-                               amdgpu_gfx_select_se_sh(adev, j, 0x0, k);
+                               amdgpu_gfx_select_se_sh(adev, j, 0x0, k, 0);
                                RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i]));
                        }
                }
@@ -6670,7 +6670,7 @@ static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
        for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) {
                for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) {
                        for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) {
-                               amdgpu_gfx_select_se_sh(adev, j, 0, k);
+                               amdgpu_gfx_select_se_sh(adev, j, 0, k, 0);
                                reg_value =
                                        RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i]));
                                if (reg_value)
@@ -6685,7 +6685,7 @@ static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev,
        err_data->ce_count += sec_count;
        err_data->ue_count += ded_count;
 
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        gfx_v9_0_query_utc_edc_status(adev, err_data);
@@ -7145,7 +7145,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
                        mask = 1;
                        ao_bitmap = 0;
                        counter = 0;
-                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff, 0);
                        gfx_v9_0_set_user_cu_inactive_bitmap(
                                adev, disable_masks[i * adev->gfx.config.max_sh_per_se + j]);
                        bitmap = gfx_v9_0_get_cu_active_bitmap(adev);
@@ -7178,7 +7178,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev,
                        cu_info->ao_cu_bitmap[i % 4][j + i / 4] = ao_bitmap;
                }
        }
-       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       amdgpu_gfx_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/amdgpu/gfx_v9_0.h b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
index dfe8d4841f582cb552bb3f09eb11c3585984ff2a..f9f6edc5e5589cfb798f724ec8531f3e46ee02bc 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.h
@@ -27,6 +27,6 @@
 extern const struct amdgpu_ip_block_version gfx_v9_0_ip_block;
 
 void gfx_v9_0_select_se_sh(struct amdgpu_device *adev, u32 se_num, u32 sh_num,
-                          u32 instance);
+                          u32 instance, int xcc_id);
 
 #endif

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
index 93438770ca1a18240d1b3037031ef2998539756b..d648a29c33e0a1cbb56363e69640d0b6d239e879 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c
@@ -777,7 +777,7 @@ void gfx_v9_4_2_set_power_brake_sequence(struct amdgpu_device *adev)
 {
        u32 tmp;
 
-       gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
 
        tmp = 0;
        tmp = REG_SET_FIELD(tmp, GC_THROTTLE_CTRL, PATTERN_MODE, 1);

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 f1c04140e71779d93690c588b824ba8caabc756c..b67be666f38a1b52d1b08d7a995c3a4995c4c78f 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -568,7 +568,8 @@ static int gfx_v9_4_3_mec_init(struct amdgpu_device *adev)
 static void gfx_v9_4_3_select_se_sh(struct amdgpu_device *adev,
                                    u32 se_num,
                                    u32 sh_num,
-                                   u32 instance)
+                                   u32 instance,
+                                   int xcc_id)
 {
        u32 data;
 
@@ -591,7 +592,7 @@ static void gfx_v9_4_3_select_se_sh(struct amdgpu_device *adev,
        else
                data = REG_SET_FIELD(data, GRBM_GFX_INDEX, SH_INDEX, sh_num);
 
-       WREG32_SOC15_RLC_SHADOW_EX(reg, GC, 0, regGRBM_GFX_INDEX, data);
+       WREG32_SOC15_RLC_SHADOW_EX(reg, GC, xcc_id, regGRBM_GFX_INDEX, data);
 }
 
 static uint32_t wave_read_ind(struct amdgpu_device *adev, uint32_t simd, uint32_t wave, uint32_t address)
@@ -907,13 +908,13 @@ static void gfx_v9_4_3_setup_rb(struct amdgpu_device *adev, int xcc_id)
        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++) {
-                       gfx_v9_4_3_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v9_4_3_select_se_sh(adev, i, j, 0xffffffff, xcc_id);
                        data = gfx_v9_4_3_get_rb_active_bitmap(adev);
                        active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) *
                                               rb_bitmap_width_per_sh);
                }
        }
-       gfx_v9_4_3_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v9_4_3_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, xcc_id);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        adev->gfx.config.backend_enable_mask = active_rbs;
@@ -1146,7 +1147,7 @@ static void gfx_v9_4_3_wait_for_rlc_serdes(struct amdgpu_device *adev,
        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++) {
-                       gfx_v9_4_3_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v9_4_3_select_se_sh(adev, i, j, 0xffffffff, xcc_id);
                        for (k = 0; k < adev->usec_timeout; k++) {
                                if (RREG32_SOC15(GC, 0, regRLC_SERDES_CU_MASTER_BUSY) == 0)
                                        break;
@@ -1154,7 +1155,8 @@ static void gfx_v9_4_3_wait_for_rlc_serdes(struct amdgpu_device *adev,
                        }
                        if (k == adev->usec_timeout) {
                                gfx_v9_4_3_select_se_sh(adev, 0xffffffff,
-                                                     0xffffffff, 0xffffffff);
+                                                       0xffffffff, 0xffffffff,
+                                                       xcc_id);
                                mutex_unlock(&adev->grbm_idx_mutex);
                                DRM_INFO("Timeout wait for RLC serdes %u,%u\n",
                                         i, j);
@@ -1162,7 +1164,7 @@ static void gfx_v9_4_3_wait_for_rlc_serdes(struct amdgpu_device *adev,
                        }
                }
        }
-       gfx_v9_4_3_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v9_4_3_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, xcc_id);
        mutex_unlock(&adev->grbm_idx_mutex);
 
        mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK |
@@ -3065,7 +3067,7 @@ static int gfx_v9_4_3_get_cu_info(struct amdgpu_device *adev,
                        mask = 1;
                        ao_bitmap = 0;
                        counter = 0;
-                       gfx_v9_4_3_select_se_sh(adev, i, j, 0xffffffff);
+                       gfx_v9_4_3_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);
@@ -3098,7 +3100,7 @@ static int gfx_v9_4_3_get_cu_info(struct amdgpu_device *adev,
                        cu_info->ao_cu_bitmap[i % 4][j + i / 4] = ao_bitmap;
                }
        }
-       gfx_v9_4_3_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+       gfx_v9_4_3_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/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c
index 47420b403871cfde1aad0662e3be8c56eac784f0..148049782f50d9386822890237e2d14af3bfb1fc 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/nv.c
+++ b/drivers/gpu/drm/amd/amdgpu/nv.c
@@ -371,12 +371,12 @@ static uint32_t nv_read_indexed_register(struct amdgpu_device *adev, u32 se_num,
 
        mutex_lock(&adev->grbm_idx_mutex);
        if (se_num != 0xffffffff || sh_num != 0xffffffff)
-               amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
        val = RREG32(reg_offset);
 
        if (se_num != 0xffffffff || sh_num != 0xffffffff)
-               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
        return val;
 }

diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c
index 7f99e130acd0663eec48e3e6149468fd2b833c45..f64b87b11b1b5aa7be6c5d7cad97b004e86b0ffc 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/si.c
+++ b/drivers/gpu/drm/amd/amdgpu/si.c
@@ -1181,12 +1181,12 @@ static uint32_t si_get_register_value(struct amdgpu_device *adev,
 
                mutex_lock(&adev->grbm_idx_mutex);
                if (se_num != 0xffffffff || sh_num != 0xffffffff)
-                       amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
                val = RREG32(reg_offset);
 
                if (se_num != 0xffffffff || sh_num != 0xffffffff)
-                       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                mutex_unlock(&adev->grbm_idx_mutex);
                return val;
        } else {

diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 4b79a8933476cff19717a4b7d3460507d0303307..4d1487a9836c6ad6ec18d2c8c36d58de7da65610 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -363,12 +363,12 @@ static uint32_t soc15_read_indexed_register(struct amdgpu_device *adev, u32 se_n
 
        mutex_lock(&adev->grbm_idx_mutex);
        if (se_num != 0xffffffff || sh_num != 0xffffffff)
-               amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
        val = RREG32(reg_offset);
 
        if (se_num != 0xffffffff || sh_num != 0xffffffff)
-               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
        return val;
 }

diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c
index 514bfc705d5abe2aea5915e471cc89aef7206fc9..6ef4be9322d9553cab634318c1ac689d5862b793 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/soc21.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc21.c
@@ -288,12 +288,12 @@ static uint32_t soc21_read_indexed_register(struct amdgpu_device *adev, u32 se_n
 
        mutex_lock(&adev->grbm_idx_mutex);
        if (se_num != 0xffffffff || sh_num != 0xffffffff)
-               amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
        val = RREG32(reg_offset);
 
        if (se_num != 0xffffffff || sh_num != 0xffffffff)
-               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+               amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
        mutex_unlock(&adev->grbm_idx_mutex);
        return val;
 }

diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index 531f173ade2d62c1f1ced0ef3a83596b181efb87..8e70581960fbb23db58a71ff2c2cc8613f7d954b 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -762,12 +762,12 @@ static uint32_t vi_get_register_value(struct amdgpu_device *adev,
 
                mutex_lock(&adev->grbm_idx_mutex);
                if (se_num != 0xffffffff || sh_num != 0xffffffff)
-                       amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, se_num, sh_num, 0xffffffff, 0);
 
                val = RREG32(reg_offset);
 
                if (se_num != 0xffffffff || sh_num != 0xffffffff)
-                       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff);
+                       amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff, 0);
                mutex_unlock(&adev->grbm_idx_mutex);
                return val;
        } else {