drm/amd/display: add DCN support for aarch64

This adds ARM64 support into the DCN. This mainly enables support
for Navi graphics cards. The dcn10 changes haven't been tested,
since I don't have the relevant hardware available, but there
is no way to conditionally disable them, so I've done them anyway.
Signed-off-by: Daniel Kolesa <daniel@octaforge.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/display/Kconfig

@@ -6,7 +6,7 @@ config DRM_AMD_DC
 	bool "AMD DC - Enable new display engine"
 	default y
 	select SND_HDA_COMPONENT if SND_HDA_CORE
-	select DRM_AMD_DC_DCN if (X86 || PPC64) && !(KCOV_INSTRUMENT_ALL && KCOV_ENABLE_COMPARISONS)
+	select DRM_AMD_DC_DCN if (X86 || PPC64 || (ARM64 && KERNEL_MODE_NEON)) && !(KCOV_INSTRUMENT_ALL && KCOV_ENABLE_COMPARISONS)
 	help
 	  Choose this option if you want to use the new display engine
 	  support for AMDGPU. This adds required support for Vega and

drivers/gpu/drm/amd/display/dc/calcs/Makefile

@@ -33,6 +33,10 @@ ifdef CONFIG_PPC64
 calcs_ccflags := -mhard-float -maltivec
 endif
 
+ifdef CONFIG_ARM64
+calcs_rcflags := -mgeneral-regs-only
+endif
+
 ifdef CONFIG_CC_IS_GCC
 ifeq ($(call cc-ifversion, -lt, 0701, y), y)
 IS_OLD_GCC = 1
@@ -53,6 +57,9 @@ endif
 CFLAGS_$(AMDDALPATH)/dc/calcs/dcn_calcs.o := $(calcs_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/calcs/dcn_calc_auto.o := $(calcs_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/calcs/dcn_calc_math.o := $(calcs_ccflags) -Wno-tautological-compare
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/calcs/dcn_calcs.o := $(calcs_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/calcs/dcn_calc_auto.o := $(calcs_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/calcs/dcn_calc_math.o := $(calcs_rcflags)
 
 BW_CALCS = dce_calcs.o bw_fixed.o custom_float.o
 

drivers/gpu/drm/amd/display/dc/clk_mgr/Makefile

@@ -104,6 +104,13 @@ ifdef CONFIG_PPC64
 CFLAGS_$(AMDDALPATH)/dc/clk_mgr/dcn21/rn_clk_mgr.o := $(call cc-option,-mno-gnu-attribute)
 endif
 
+# prevent build errors:
+# ...: '-mgeneral-regs-only' is incompatible with the use of floating-point types
+# this file is unused on arm64, just like on ppc64
+ifdef CONFIG_ARM64
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/clk_mgr/dcn21/rn_clk_mgr.o := -mgeneral-regs-only
+endif
+
 AMD_DAL_CLK_MGR_DCN21 = $(addprefix $(AMDDALPATH)/dc/clk_mgr/dcn21/,$(CLK_MGR_DCN21))
 
 AMD_DISPLAY_FILES += $(AMD_DAL_CLK_MGR_DCN21)

drivers/gpu/drm/amd/display/dc/dcn10/Makefile

@@ -31,4 +31,11 @@ DCN10 = dcn10_init.o dcn10_resource.o dcn10_ipp.o dcn10_hw_sequencer.o \
 
 AMD_DAL_DCN10 = $(addprefix $(AMDDALPATH)/dc/dcn10/,$(DCN10))
 
+# fix:
+# ...: '-mgeneral-regs-only' is incompatible with the use of floating-point types
+# aarch64 does not support soft-float, so use hard-float and handle this in code
+ifdef CONFIG_ARM64
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dcn10/dcn10_resource.o := -mgeneral-regs-only
+endif
+
 AMD_DISPLAY_FILES += $(AMD_DAL_DCN10)

drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c

@@ -1331,6 +1331,47 @@ static uint32_t read_pipe_fuses(struct dc_context *ctx)
 	return value;
 }
 
+/*
+ * Some architectures don't support soft-float (e.g. aarch64), on those
+ * this function has to be called with hardfloat enabled, make sure not
+ * to inline it so whatever fp stuff is done stays inside
+ */
+static noinline void dcn10_resource_construct_fp(
+	struct dc *dc)
+{
+	if (dc->ctx->dce_version == DCN_VERSION_1_01) {
+		struct dcn_soc_bounding_box *dcn_soc = dc->dcn_soc;
+		struct dcn_ip_params *dcn_ip = dc->dcn_ip;
+		struct display_mode_lib *dml = &dc->dml;
+
+		dml->ip.max_num_dpp = 3;
+		/* TODO how to handle 23.84? */
+		dcn_soc->dram_clock_change_latency = 23;
+		dcn_ip->max_num_dpp = 3;
+	}
+	if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) {
+		dc->dcn_soc->urgent_latency = 3;
+		dc->debug.disable_dmcu = true;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 41.60f;
+	}
+
+
+	dc->dcn_soc->number_of_channels = dc->ctx->asic_id.vram_width / ddr4_dram_width;
+	ASSERT(dc->dcn_soc->number_of_channels < 3);
+	if (dc->dcn_soc->number_of_channels == 0)/*old sbios bug*/
+		dc->dcn_soc->number_of_channels = 2;
+
+	if (dc->dcn_soc->number_of_channels == 1) {
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 19.2f;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = 17.066f;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = 14.933f;
+		dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = 12.8f;
+		if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) {
+			dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 20.80f;
+		}
+	}
+}
+
 static bool dcn10_resource_construct(
 	uint8_t num_virtual_links,
 	struct dc *dc,
@@ -1482,37 +1523,15 @@ static bool dcn10_resource_construct(
 	memcpy(dc->dcn_ip, &dcn10_ip_defaults, sizeof(dcn10_ip_defaults));
 	memcpy(dc->dcn_soc, &dcn10_soc_defaults, sizeof(dcn10_soc_defaults));
 
-	if (dc->ctx->dce_version == DCN_VERSION_1_01) {
-		struct dcn_soc_bounding_box *dcn_soc = dc->dcn_soc;
-		struct dcn_ip_params *dcn_ip = dc->dcn_ip;
-		struct display_mode_lib *dml = &dc->dml;
-
-		dml->ip.max_num_dpp = 3;
-		/* TODO how to handle 23.84? */
-		dcn_soc->dram_clock_change_latency = 23;
-		dcn_ip->max_num_dpp = 3;
-	}
-	if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) {
-		dc->dcn_soc->urgent_latency = 3;
-		dc->debug.disable_dmcu = true;
-		dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 41.60f;
-	}
-
-
-	dc->dcn_soc->number_of_channels = dc->ctx->asic_id.vram_width / ddr4_dram_width;
-	ASSERT(dc->dcn_soc->number_of_channels < 3);
-	if (dc->dcn_soc->number_of_channels == 0)/*old sbios bug*/
-		dc->dcn_soc->number_of_channels = 2;
-
-	if (dc->dcn_soc->number_of_channels == 1) {
-		dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 19.2f;
-		dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = 17.066f;
-		dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = 14.933f;
-		dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = 12.8f;
-		if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) {
-			dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 20.80f;
-		}
-	}
+#if defined(CONFIG_ARM64)
+	/* Aarch64 does not support -msoft-float/-mfloat-abi=soft */
+	DC_FP_START();
+	dcn10_resource_construct_fp(dc);
+	DC_FP_END();
+#else
+	/* Other architectures we build for build this with soft-float */
+	dcn10_resource_construct_fp(dc);
+#endif
 
 	pool->base.pp_smu = dcn10_pp_smu_create(ctx);
 

drivers/gpu/drm/amd/display/dc/dcn20/Makefile

@@ -17,6 +17,10 @@ ifdef CONFIG_PPC64
 CFLAGS_$(AMDDALPATH)/dc/dcn20/dcn20_resource.o := -mhard-float -maltivec
 endif
 
+ifdef CONFIG_ARM64
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dcn20/dcn20_resource.o := -mgeneral-regs-only
+endif
+
 ifdef CONFIG_CC_IS_GCC
 ifeq ($(call cc-ifversion, -lt, 0701, y), y)
 IS_OLD_GCC = 1

drivers/gpu/drm/amd/display/dc/dcn21/Makefile

@@ -13,6 +13,10 @@ ifdef CONFIG_PPC64
 CFLAGS_$(AMDDALPATH)/dc/dcn21/dcn21_resource.o := -mhard-float -maltivec
 endif
 
+ifdef CONFIG_ARM64
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dcn21/dcn21_resource.o := -mgeneral-regs-only
+endif
+
 ifdef CONFIG_CC_IS_GCC
 ifeq ($(call cc-ifversion, -lt, 0701, y), y)
 IS_OLD_GCC = 1

drivers/gpu/drm/amd/display/dc/dml/Makefile

@@ -33,6 +33,10 @@ ifdef CONFIG_PPC64
 dml_ccflags := -mhard-float -maltivec
 endif
 
+ifdef CONFIG_ARM64
+dml_rcflags := -mgeneral-regs-only
+endif
+
 ifdef CONFIG_CC_IS_GCC
 ifeq ($(call cc-ifversion, -lt, 0701, y), y)
 IS_OLD_GCC = 1
@@ -60,6 +64,13 @@ CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20v2.o := $(dml_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20v2.o := $(dml_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/dml/dcn21/display_mode_vba_21.o := $(dml_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/dml/dcn21/display_rq_dlg_calc_21.o := $(dml_ccflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_mode_vba.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20v2.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20v2.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn21/display_mode_vba_21.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn21/display_rq_dlg_calc_21.o := $(dml_rcflags)
 endif
 ifdef CONFIG_DRM_AMD_DC_DCN3_0
 CFLAGS_$(AMDDALPATH)/dc/dml/dcn30/display_mode_vba_30.o := $(dml_ccflags) -Wframe-larger-than=2048
@@ -67,6 +78,8 @@ CFLAGS_$(AMDDALPATH)/dc/dml/dcn30/display_rq_dlg_calc_30.o := $(dml_ccflags)
 endif
 CFLAGS_$(AMDDALPATH)/dc/dml/dml1_display_rq_dlg_calc.o := $(dml_ccflags)
 CFLAGS_$(AMDDALPATH)/dc/dml/display_rq_dlg_helpers.o := $(dml_ccflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dml1_display_rq_dlg_calc.o := $(dml_rcflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_rq_dlg_helpers.o := $(dml_rcflags)
 
 DML = display_mode_lib.o display_rq_dlg_helpers.o dml1_display_rq_dlg_calc.o \
 

drivers/gpu/drm/amd/display/dc/dsc/Makefile

@@ -10,6 +10,10 @@ ifdef CONFIG_PPC64
 dsc_ccflags := -mhard-float -maltivec
 endif
 
+ifdef CONFIG_ARM64
+dsc_rcflags := -mgeneral-regs-only
+endif
+
 ifdef CONFIG_CC_IS_GCC
 ifeq ($(call cc-ifversion, -lt, 0701, y), y)
 IS_OLD_GCC = 1
@@ -28,6 +32,7 @@ endif
 endif
 
 CFLAGS_$(AMDDALPATH)/dc/dsc/rc_calc.o := $(dsc_ccflags)
+CFLAGS_REMOVE_$(AMDDALPATH)/dc/dsc/rc_calc.o := $(dsc_rcflags)
 
 DSC = dc_dsc.o rc_calc.o rc_calc_dpi.o
 

drivers/gpu/drm/amd/display/dc/os_types.h

@@ -55,6 +55,10 @@
 #include <asm/fpu/api.h>
 #define DC_FP_START() kernel_fpu_begin()
 #define DC_FP_END() kernel_fpu_end()
+#elif defined(CONFIG_ARM64)
+#include <asm/neon.h>
+#define DC_FP_START() kernel_neon_begin()
+#define DC_FP_END() kernel_neon_end()
 #elif defined(CONFIG_PPC64)
 #include <asm/switch_to.h>
 #include <asm/cputable.h>