drm/amdgpu: convert IP version array to include instances

Allow us to query instances versions more cleanly.

Instancing support is not consistent unfortunately. SDMA is a
good example.  Sienna cichlid has 4 total SDMA instances, each
enumerated separately (HWIDs 42, 43, 68, 69).  Arcturus has 8
total SDMA instances, but they are enumerated as multiple
instances of the same HWIDs (4x HWID 42, 4x HWID 43).  UMC
is another example.  On most chips there are multiple
instances with the same HWID.  This allows us to support both
forms.

v2: rebase
v3: clarify instancing support

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 40 insertions, 40 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
index 4c632306ed70..71bb3c0dc1da 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
@@ -1537,7 +1537,7 @@ static u32 gfx_v10_rlcg_rw(struct amdgpu_device *adev, u32 offset, u32 v, uint32
 	scratch_reg3 = adev->rmmio +
 		       (adev->reg_offset[GC_HWIP][0][mmSCRATCH_REG1_BASE_IDX] + mmSCRATCH_REG3) * 4;
 
-	if (adev->ip_versions[GC_HWIP] >= IP_VERSION(10, 3, 0)) {
+	if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) {
 		spare_int = adev->rmmio +
 			    (adev->reg_offset[GC_HWIP][0][mmRLC_SPARE_INT_0_Sienna_Cichlid_BASE_IDX]
 			     + mmRLC_SPARE_INT_0_Sienna_Cichlid) * 4;
@@ -3727,7 +3727,7 @@ static void gfx_v10_0_set_kiq_pm4_funcs(struct amdgpu_device *adev)
 
 static void gfx_v10_0_init_spm_golden_registers(struct amdgpu_device *adev)
 {
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 		soc15_program_register_sequence(adev,
 						golden_settings_gc_rlc_spm_10_0_nv10,
@@ -3750,7 +3750,7 @@ static void gfx_v10_0_init_spm_golden_registers(struct amdgpu_device *adev)
 
 static void gfx_v10_0_init_golden_registers(struct amdgpu_device *adev)
 {
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 		soc15_program_register_sequence(adev,
 						golden_settings_gc_10_1,
@@ -3985,7 +3985,7 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev)
 {
 	adev->gfx.cp_fw_write_wait = false;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 2):
 	case IP_VERSION(10, 1, 1):
@@ -4066,7 +4066,7 @@ static bool gfx_v10_0_navi10_gfxoff_should_enable(struct amdgpu_device *adev)
 
 static void gfx_v10_0_check_gfxoff_flag(struct amdgpu_device *adev)
 {
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 		if (!gfx_v10_0_navi10_gfxoff_should_enable(adev))
 			adev->pm.pp_feature &= ~PP_GFXOFF_MASK;
@@ -4093,7 +4093,7 @@ static int gfx_v10_0_init_microcode(struct amdgpu_device *adev)
 
 	DRM_DEBUG("\n");
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 		chip_name = "navi10";
 		break;
@@ -4684,7 +4684,7 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev)
 
 	adev->gfx.funcs = &gfx_v10_0_gfx_funcs;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 1):
 	case IP_VERSION(10, 1, 2):
@@ -4818,7 +4818,7 @@ static int gfx_v10_0_sw_init(void *handle)
 	struct amdgpu_kiq *kiq;
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 1):
 	case IP_VERSION(10, 1, 2):
@@ -5068,8 +5068,8 @@ static void gfx_v10_0_setup_rb(struct amdgpu_device *adev)
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
 			bitmap = i * adev->gfx.config.max_sh_per_se + j;
-			if (((adev->ip_versions[GC_HWIP] == IP_VERSION(10, 3, 0)) ||
-				(adev->ip_versions[GC_HWIP] == IP_VERSION(10, 3, 3))) &&
+			if (((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) ||
+				(adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3))) &&
 			    ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1))
 				continue;
 			gfx_v10_0_select_se_sh(adev, i, j, 0xffffffff);
@@ -5096,7 +5096,7 @@ static u32 gfx_v10_0_init_pa_sc_tile_steering_override(struct amdgpu_device *ade
 
 	/* for ASICs that integrates GFX v10.3
 	 * pa_sc_tile_steering_override should be set to 0 */
-	if (adev->ip_versions[GC_HWIP] >= IP_VERSION(10, 3, 0))
+	if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0))
 		return 0;
 
 	/* init num_sc */
@@ -5249,7 +5249,7 @@ static void gfx_v10_0_get_tcc_info(struct amdgpu_device *adev)
 	/* TCCs are global (not instanced). */
 	uint32_t tcc_disable;
 
-	if (adev->ip_versions[GC_HWIP] >= IP_VERSION(10, 3, 0)) {
+	if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) {
 		tcc_disable = RREG32_SOC15(GC, 0, mmCGTS_TCC_DISABLE_gc_10_3) |
 			      RREG32_SOC15(GC, 0, mmCGTS_USER_TCC_DISABLE_gc_10_3);
 	} else {
@@ -5326,7 +5326,7 @@ static int gfx_v10_0_init_csb(struct amdgpu_device *adev)
 	adev->gfx.rlc.funcs->get_csb_buffer(adev, adev->gfx.rlc.cs_ptr);
 
 	/* csib */
-	if (adev->ip_versions[GC_HWIP] == IP_VERSION(10, 1, 2)) {
+	if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) {
 		WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_ADDR_HI,
 				adev->gfx.rlc.clear_state_gpu_addr >> 32);
 		WREG32_SOC15_RLC(GC, 0, mmRLC_CSIB_ADDR_LO,
@@ -5948,7 +5948,7 @@ static int gfx_v10_0_cp_gfx_enable(struct amdgpu_device *adev, bool enable)
 	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, PFP_HALT, enable ? 0 : 1);
 	tmp = REG_SET_FIELD(tmp, CP_ME_CNTL, CE_HALT, enable ? 0 : 1);
 
-	if (adev->ip_versions[GC_HWIP] == IP_VERSION(10, 1, 2)) {
+	if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) {
 		WREG32_SOC15_RLC(GC, 0, mmCP_ME_CNTL, tmp);
 	} else {
 		WREG32_SOC15(GC, 0, mmCP_ME_CNTL, tmp);
@@ -6337,7 +6337,7 @@ static void gfx_v10_0_cp_gfx_set_doorbell(struct amdgpu_device *adev,
 		}
 		WREG32_SOC15(GC, 0, mmCP_RB_DOORBELL_CONTROL, tmp);
 	}
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 1):
@@ -6474,7 +6474,7 @@ static int gfx_v10_0_cp_gfx_resume(struct amdgpu_device *adev)
 static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 {
 	if (enable) {
-		switch (adev->ip_versions[GC_HWIP]) {
+		switch (adev->ip_versions[GC_HWIP][0]) {
 		case IP_VERSION(10, 3, 0):
 		case IP_VERSION(10, 3, 2):
 		case IP_VERSION(10, 3, 1):
@@ -6488,7 +6488,7 @@ static void gfx_v10_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
 			break;
 		}
 	} else {
-		switch (adev->ip_versions[GC_HWIP]) {
+		switch (adev->ip_versions[GC_HWIP][0]) {
 		case IP_VERSION(10, 3, 0):
 		case IP_VERSION(10, 3, 2):
 		case IP_VERSION(10, 3, 1):
@@ -6586,7 +6586,7 @@ static void gfx_v10_0_kiq_setting(struct amdgpu_ring *ring)
 	struct amdgpu_device *adev = ring->adev;
 
 	/* tell RLC which is KIQ queue */
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 1):
@@ -7303,7 +7303,7 @@ static bool gfx_v10_0_check_grbm_cam_remapping(struct amdgpu_device *adev)
 
 	/* check if mmVGT_ESGS_RING_SIZE_UMD
 	 * has been remapped to mmVGT_ESGS_RING_SIZE */
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 4):
@@ -7350,7 +7350,7 @@ static void gfx_v10_0_setup_grbm_cam_remapping(struct amdgpu_device *adev)
 	 * index will auto-inc after each data writting */
 	WREG32_SOC15(GC, 0, mmGRBM_CAM_INDEX, 0);
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 1):
@@ -7520,19 +7520,19 @@ static int gfx_v10_0_hw_init(void *handle)
 	 * init golden registers and rlc resume may override some registers,
 	 * reconfig them here
 	 */
-	if (adev->ip_versions[GC_HWIP] == IP_VERSION(10, 1, 10) ||
-	    adev->ip_versions[GC_HWIP] == IP_VERSION(10, 1, 1) ||
-	    adev->ip_versions[GC_HWIP] == IP_VERSION(10, 1, 2))
+	if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 10) ||
+	    adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 1) ||
+	    adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2))
 		gfx_v10_0_tcp_harvest(adev);
 
 	r = gfx_v10_0_cp_resume(adev);
 	if (r)
 		return r;
 
-	if (adev->ip_versions[GC_HWIP] == IP_VERSION(10, 3, 0))
+	if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0))
 		gfx_v10_3_program_pbb_mode(adev);
 
-	if (adev->ip_versions[GC_HWIP] >= IP_VERSION(10, 3, 0))
+	if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0))
 		gfx_v10_3_set_power_brake_sequence(adev);
 
 	return r;
@@ -7584,7 +7584,7 @@ static int gfx_v10_0_hw_fini(void *handle)
 	if (amdgpu_sriov_vf(adev)) {
 		gfx_v10_0_cp_gfx_enable(adev, false);
 		/* Program KIQ position of RLC_CP_SCHEDULERS during destroy */
-		if (adev->ip_versions[GC_HWIP] >= IP_VERSION(10, 3, 0)) {
+		if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 3, 0)) {
 			tmp = RREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid);
 			tmp &= 0xffffff00;
 			WREG32_SOC15(GC, 0, mmRLC_CP_SCHEDULERS_Sienna_Cichlid, tmp);
@@ -7670,7 +7670,7 @@ static int gfx_v10_0_soft_reset(void *handle)
 
 	/* GRBM_STATUS2 */
 	tmp = RREG32_SOC15(GC, 0, mmGRBM_STATUS2);
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 1):
@@ -7726,7 +7726,7 @@ static uint64_t gfx_v10_0_get_gpu_clock_counter(struct amdgpu_device *adev)
 {
 	uint64_t clock, clock_lo, clock_hi, hi_check;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 1):
 	case IP_VERSION(10, 3, 3):
 		clock = (uint64_t)RREG32_SOC15(SMUIO, 0, mmGOLDEN_TSC_COUNT_LOWER_Vangogh) |
@@ -7784,7 +7784,7 @@ static int gfx_v10_0_early_init(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 1):
 	case IP_VERSION(10, 1, 2):
@@ -7848,7 +7848,7 @@ static void gfx_v10_0_set_safe_mode(struct amdgpu_device *adev)
 	data = RLC_SAFE_MODE__CMD_MASK;
 	data |= (1 << RLC_SAFE_MODE__MESSAGE__SHIFT);
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 1):
@@ -7884,7 +7884,7 @@ static void gfx_v10_0_unset_safe_mode(struct amdgpu_device *adev)
 	uint32_t data;
 
 	data = RLC_SAFE_MODE__CMD_MASK;
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 3, 0):
 	case IP_VERSION(10, 3, 2):
 	case IP_VERSION(10, 3, 1):
@@ -8193,7 +8193,7 @@ static void gfx_v10_0_apply_medium_grain_clock_gating_workaround(struct amdgpu_d
 		mmCGTS_SA1_QUAD1_SM_CTRL_REG
 	};
 
-	if (adev->ip_versions[GC_HWIP] == IP_VERSION(10, 1, 2)) {
+	if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 1, 2)) {
 		for (i = 0; i < ARRAY_SIZE(tcp_ctrl_regs_nv12); i++) {
 			reg_idx = adev->reg_offset[GC_HWIP][0][mmCGTS_SA0_WGP00_CU0_TCP_CTRL_REG_BASE_IDX] +
 				  tcp_ctrl_regs_nv12[i];
@@ -8238,8 +8238,8 @@ static int gfx_v10_0_update_gfx_clock_gating(struct amdgpu_device *adev,
 		/* ===  CGCG + CGLS === */
 		gfx_v10_0_update_coarse_grain_clock_gating(adev, enable);
 
-		if ((adev->ip_versions[GC_HWIP] >= IP_VERSION(10, 1, 10)) &&
-		     (adev->ip_versions[GC_HWIP] <= IP_VERSION(10, 1, 2)))
+		if ((adev->ip_versions[GC_HWIP][0] >= IP_VERSION(10, 1, 10)) &&
+		     (adev->ip_versions[GC_HWIP][0] <= IP_VERSION(10, 1, 2)))
 			gfx_v10_0_apply_medium_grain_clock_gating_workaround(adev);
 	} else {
 		/* CGCG/CGLS should be disabled before MGCG/MGLS
@@ -8335,7 +8335,7 @@ static void gfx_v10_cntl_power_gating(struct amdgpu_device *adev, bool enable)
 	 * Power/performance team will optimize it and might give a new value later.
 	 */
 	if (enable && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) {
-		switch (adev->ip_versions[GC_HWIP]) {
+		switch (adev->ip_versions[GC_HWIP][0]) {
 		case IP_VERSION(10, 3, 1):
 			data = 0x4E20 & RLC_PG_DELAY_3__CGCG_ACTIVE_BEFORE_CGPG_MASK_Vangogh;
 			WREG32_SOC15(GC, 0, mmRLC_PG_DELAY_3, data);
@@ -8399,7 +8399,7 @@ static int gfx_v10_0_set_powergating_state(void *handle,
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 1):
 	case IP_VERSION(10, 1, 2):
@@ -8428,7 +8428,7 @@ static int gfx_v10_0_set_clockgating_state(void *handle,
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 1):
 	case IP_VERSION(10, 1, 2):
@@ -9541,7 +9541,7 @@ static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev)
 
 static void gfx_v10_0_set_rlc_funcs(struct amdgpu_device *adev)
 {
-	switch (adev->ip_versions[GC_HWIP]) {
+	switch (adev->ip_versions[GC_HWIP][0]) {
 	case IP_VERSION(10, 1, 10):
 	case IP_VERSION(10, 1, 1):
 	case IP_VERSION(10, 1, 3):
@@ -9641,8 +9641,8 @@ static int gfx_v10_0_get_cu_info(struct amdgpu_device *adev,
 	for (i = 0; i < adev->gfx.config.max_shader_engines; i++) {
 		for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) {
 			bitmap = i * adev->gfx.config.max_sh_per_se + j;
-			if (((adev->ip_versions[GC_HWIP] == IP_VERSION(10, 3, 0)) ||
-				(adev->ip_versions[GC_HWIP] == IP_VERSION(10, 3, 3))) &&
+			if (((adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 0)) ||
+				(adev->ip_versions[GC_HWIP][0] == IP_VERSION(10, 3, 3))) &&
 			    ((gfx_v10_3_get_disabled_sa(adev) >> bitmap) & 1))
 				continue;
 			mask = 1;