drm/amdgpu/gfx: Add mmCOMPUTE_STATIC_THREAD_MGMT_SE4-7 to support Arcturus

Add mmCOMPUTE_STATIC_THREAD_MGMT_SE4-7 to support Arcturus

Signed-off-by: James Zhu <James.Zhu@amd.com>
Reviewed-by: Yong Zhao <Yong.Zhao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index f9d2289..489e467 100644 (file)
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -3927,42 +3927,54 @@ static const u32 sgpr_init_compute_shader[] =
 };
 
 static const struct soc15_reg_entry vgpr_init_regs[] = {
-   { 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_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), 0x3f },
    { 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_STATIC_THREAD_MGMT_SE0), 0x000000ff },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x000000ff },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x000000ff },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x000000ff },
    { 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), 8 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x000000ff },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x000000ff },
 };
 
 static const struct soc15_reg_entry sgpr2_init_regs[] = {
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x0000ff00 },
-   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x0000ff00 },
    { 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), 8 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 },
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */
    { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x0000ff00 },
+   { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x0000ff00 },
 };
 
 static const struct soc15_reg_entry sec_ded_counter_registers[] = {
@@ -4064,7 +4076,7 @@ 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 = ARRAY_SIZE(vgpr_init_regs);
+       int gpr_reg_size = compute_dim_x / 16 + 6;
        int sec_ded_counter_reg_size = ARRAY_SIZE(sec_ded_counter_registers);
 
        /* only support when RAS is enabled */