[PATCH] drm/amd/powerplay: bug fix for pcie parameters override

Kenneth Feng kenneth.feng at amd.com
Fri Oct 11 10:09:08 UTC 2019


Bug fix for pcie paramerers override on swsmu.
Below is a scenario to have this problem.
pptable definition on pcie dpm:
0 -> pcie gen speed:1, pcie lanes: *16
1 -> pcie gen speed:4, pcie lanes: *16
Then if we have a system only have the capbility:
pcie gen speed: 3, pcie lanes: *8,
we will override dpm 1 to pcie gen speed 3, pcie lanes *8.
But the code skips the dpm 0 configuration.
So the real pcie dpm parameters are:
0 -> pcie gen speed:1, pcie lanes: *16
1 -> pcie gen speed:3, pcie lanes: *8
Then the wrong pcie lanes will be toggled.

Signed-off-by: Kenneth Feng <kenneth.feng at amd.com>
---
 drivers/gpu/drm/amd/powerplay/amdgpu_smu.c      |   44 -
 drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h  |    8 +
 drivers/gpu/drm/amd/powerplay/navi10_ppt.c      |   23 +
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c       |   44 +
 drivers/gpu/drm/amd/powerplay/smu_v11_0.c_mi100 | 1404 +++++++++++++++++++++++
 drivers/gpu/drm/amd/powerplay/vega20_ppt.c      |   25 +-
 6 files changed, 1503 insertions(+), 45 deletions(-)
 create mode 100644 drivers/gpu/drm/amd/powerplay/smu_v11_0.c_mi100

diff --git a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
index c9266ea..de54da2 100644
--- a/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
+++ b/drivers/gpu/drm/amd/powerplay/amdgpu_smu.c
@@ -945,50 +945,6 @@ static int smu_fini_fb_allocations(struct smu_context *smu)
 	return 0;
 }
 
-static int smu_override_pcie_parameters(struct smu_context *smu)
-{
-	struct amdgpu_device *adev = smu->adev;
-	uint32_t pcie_gen = 0, pcie_width = 0, smu_pcie_arg;
-	int ret;
-
-	if (adev->flags & AMD_IS_APU)
-		return 0;
-
-	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4)
-		pcie_gen = 3;
-	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)
-		pcie_gen = 2;
-	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2)
-		pcie_gen = 1;
-	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1)
-		pcie_gen = 0;
-
-	/* Bit 31:16: LCLK DPM level. 0 is DPM0, and 1 is DPM1
-	 * Bit 15:8:  PCIE GEN, 0 to 3 corresponds to GEN1 to GEN4
-	 * Bit 7:0:   PCIE lane width, 1 to 7 corresponds is x1 to x32
-	 */
-	if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X16)
-		pcie_width = 6;
-	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X12)
-		pcie_width = 5;
-	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X8)
-		pcie_width = 4;
-	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X4)
-		pcie_width = 3;
-	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X2)
-		pcie_width = 2;
-	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X1)
-		pcie_width = 1;
-
-	smu_pcie_arg = (1 << 16) | (pcie_gen << 8) | pcie_width;
-	ret = smu_send_smc_msg_with_param(smu,
-					  SMU_MSG_OverridePcieParameters,
-					  smu_pcie_arg);
-	if (ret)
-		pr_err("[%s] Attempt to override pcie params failed!\n", __func__);
-	return ret;
-}
-
 static int smu_smc_table_hw_init(struct smu_context *smu,
 				 bool initialize)
 {
diff --git a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
index ccf711c..809de0d 100644
--- a/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
+++ b/drivers/gpu/drm/amd/powerplay/inc/amdgpu_smu.h
@@ -468,6 +468,7 @@ struct pptable_funcs {
 	int (*get_power_limit)(struct smu_context *smu, uint32_t *limit, bool asic_default);
 	int (*get_dpm_clk_limited)(struct smu_context *smu, enum smu_clk_type clk_type,
 				   uint32_t dpm_level, uint32_t *freq);
+	int (*update_pcie_parameters)(struct smu_context *smu, uint32_t pcie_gen_cap, uint32_t pcie_width_cap);
 };
 
 struct smu_funcs
@@ -550,6 +551,7 @@ struct smu_funcs
 	int (*mode2_reset)(struct smu_context *smu);
 	int (*get_dpm_ultimate_freq)(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t *min, uint32_t *max);
 	int (*set_soft_freq_limited_range)(struct smu_context *smu, enum smu_clk_type clk_type, uint32_t min, uint32_t max);
+	int (*override_pcie_parameters)(struct smu_context *smu);
 };
 
 #define smu_init_microcode(smu) \
@@ -782,6 +784,12 @@ struct smu_funcs
 #define smu_set_soft_freq_limited_range(smu, clk_type, min, max) \
 		((smu)->funcs->set_soft_freq_limited_range ? (smu)->funcs->set_soft_freq_limited_range((smu), (clk_type), (min), (max)) : -EINVAL)
 
+#define smu_override_pcie_parameters(smu) \
+		((smu)->funcs->override_pcie_parameters ? (smu)->funcs->override_pcie_parameters((smu)) : 0)
+
+#define smu_update_pcie_parameters(smu, pcie_gen_cap, pcie_width_cap) \
+		((smu)->ppt_funcs->update_pcie_parameters ? (smu)->ppt_funcs->update_pcie_parameters((smu), (pcie_gen_cap), (pcie_width_cap)) : 0)
+
 extern int smu_get_atom_data_table(struct smu_context *smu, uint32_t table,
 				   uint16_t *size, uint8_t *frev, uint8_t *crev,
 				   uint8_t **addr);
diff --git a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
index a583cf8..a2f33cf 100644
--- a/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/navi10_ppt.c
@@ -1592,6 +1592,28 @@ static int navi10_get_power_limit(struct smu_context *smu,
 	return 0;
 }
 
+static int navi10_update_pcie_parameters(struct smu_context *smu,
+				     uint32_t pcie_gen_cap,
+				     uint32_t pcie_width_cap)
+{
+	PPTable_t *pptable = smu->smu_table.driver_pptable;
+	int ret, i;
+	uint32_t smu_pcie_arg;
+
+	for (i = 0; i < NUM_LINK_LEVELS; i++) {
+		smu_pcie_arg = (i << 16) |
+			((pptable->PcieGenSpeed[i] <= pcie_gen_cap) ? (pptable->PcieGenSpeed[i] << 8) :
+				(pcie_gen_cap << 8)) | ((pptable->PcieLaneCount[i] <= pcie_width_cap) ?
+					pptable->PcieLaneCount[i] : pcie_width_cap);
+		ret = smu_send_smc_msg_with_param(smu,
+					  SMU_MSG_OverridePcieParameters,
+					  smu_pcie_arg);
+	}
+
+	return ret;
+}
+
+
 static const struct pptable_funcs navi10_ppt_funcs = {
 	.tables_init = navi10_tables_init,
 	.alloc_dpm_context = navi10_allocate_dpm_context,
@@ -1630,6 +1652,7 @@ static const struct pptable_funcs navi10_ppt_funcs = {
 	.get_thermal_temperature_range = navi10_get_thermal_temperature_range,
 	.display_disable_memory_clock_switch = navi10_display_disable_memory_clock_switch,
 	.get_power_limit = navi10_get_power_limit,
+	.update_pcie_parameters = navi10_update_pcie_parameters,
 };
 
 void navi10_set_ppt_funcs(struct smu_context *smu)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
index c9e90d5..a812ae5 100644
--- a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c
@@ -35,6 +35,7 @@
 #include "vega20_ppt.h"
 #include "arcturus_ppt.h"
 #include "navi10_ppt.h"
+#include "amd_pcie.h"
 
 #include "asic_reg/thm/thm_11_0_2_offset.h"
 #include "asic_reg/thm/thm_11_0_2_sh_mask.h"
@@ -1792,6 +1793,48 @@ static int smu_v11_0_set_soft_freq_limited_range(struct smu_context *smu, enum s
 	return ret;
 }
 
+static int smu_v11_0_override_pcie_parameters(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	uint32_t pcie_gen = 0, pcie_width = 0;
+	int ret;
+
+	if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4)
+		pcie_gen = 3;
+	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3)
+		pcie_gen = 2;
+	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2)
+		pcie_gen = 1;
+	else if (adev->pm.pcie_gen_mask & CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1)
+		pcie_gen = 0;
+
+	/* Bit 31:16: LCLK DPM level. 0 is DPM0, and 1 is DPM1
+	 * Bit 15:8:  PCIE GEN, 0 to 3 corresponds to GEN1 to GEN4
+	 * Bit 7:0:   PCIE lane width, 1 to 7 corresponds is x1 to x32
+	 */
+	if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X16)
+		pcie_width = 6;
+	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X12)
+		pcie_width = 5;
+	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X8)
+		pcie_width = 4;
+	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X4)
+		pcie_width = 3;
+	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X2)
+		pcie_width = 2;
+	else if (adev->pm.pcie_mlw_mask & CAIL_PCIE_LINK_WIDTH_SUPPORT_X1)
+		pcie_width = 1;
+
+	ret = smu_update_pcie_parameters(smu, pcie_gen, pcie_width);
+
+	if (ret)
+		pr_err("[%s] Attempt to override pcie params failed!\n", __func__);
+
+	return ret;
+
+}
+
+
 static const struct smu_funcs smu_v11_0_funcs = {
 	.init_microcode = smu_v11_0_init_microcode,
 	.load_microcode = smu_v11_0_load_microcode,
@@ -1844,6 +1887,7 @@ static const struct smu_funcs smu_v11_0_funcs = {
 	.baco_reset = smu_v11_0_baco_reset,
 	.get_dpm_ultimate_freq = smu_v11_0_get_dpm_ultimate_freq,
 	.set_soft_freq_limited_range = smu_v11_0_set_soft_freq_limited_range,
+	.override_pcie_parameters = smu_v11_0_override_pcie_parameters,
 };
 
 void smu_v11_0_set_smu_funcs(struct smu_context *smu)
diff --git a/drivers/gpu/drm/amd/powerplay/smu_v11_0.c_mi100 b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c_mi100
new file mode 100644
index 0000000..2e7787e
--- /dev/null
+++ b/drivers/gpu/drm/amd/powerplay/smu_v11_0.c_mi100
@@ -0,0 +1,1404 @@
+/*
+ * Copyright 2019 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "pp_debug.h"
+#include <linux/firmware.h>
+#include "amdgpu.h"
+#include "amdgpu_smu.h"
+#include "atomfirmware.h"
+#include "amdgpu_atomfirmware.h"
+#include "smu_v11_0.h"
+#include "smu11_driver_if_arcturus.h"
+#include "soc15_common.h"
+#include "atom.h"
+#include "vega20_ppt.h"
+#include "arcturus_ppt.h"
+#include "pp_thermal.h"
+
+#include "asic_reg/thm/thm_11_0_2_offset.h"
+#include "asic_reg/thm/thm_11_0_2_sh_mask.h"
+#include "asic_reg/mp/mp_9_0_offset.h"
+#include "asic_reg/mp/mp_9_0_sh_mask.h"
+#include "asic_reg/nbio/nbio_7_4_offset.h"
+#include "asic_reg/smuio/smuio_9_0_offset.h"
+#include "asic_reg/smuio/smuio_9_0_sh_mask.h"
+
+MODULE_FIRMWARE("amdgpu/vega20_smc.bin");
+MODULE_FIRMWARE("amdgpu/arcturus_smc.bin");
+
+#define SMU11_TOOL_SIZE		0x19000
+#define SMU11_THERMAL_MINIMUM_ALERT_TEMP      0
+#define SMU11_THERMAL_MAXIMUM_ALERT_TEMP      255
+
+#define SMU11_TEMPERATURE_UNITS_PER_CENTIGRADES 1000
+#define SMU11_VOLTAGE_SCALE 4
+
+#define SMC_DPM_FEATURE (FEATURE_DPM_PREFETCHER_MASK | \
+			 FEATURE_DPM_GFXCLK_MASK | \
+			 FEATURE_DPM_UCLK_MASK | \
+			 FEATURE_DPM_SOCCLK_MASK | \
+			 FEATURE_DPM_MP0CLK_MASK)
+
+static int smu_v11_0_send_msg_without_waiting(struct smu_context *smu,
+					      uint16_t msg)
+{
+	struct amdgpu_device *adev = smu->adev;
+	WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_66, msg);
+	return 0;
+}
+
+static int smu_v11_0_read_arg(struct smu_context *smu, uint32_t *arg)
+{
+	struct amdgpu_device *adev = smu->adev;
+
+	*arg = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82);
+	return 0;
+}
+
+static int smu_v11_0_wait_for_response(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	uint32_t cur_value, i;
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		cur_value = RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90);
+		if ((cur_value & MP1_C2PMSG_90__CONTENT_MASK) != 0)
+			break;
+		udelay(1);
+	}
+
+	/* timeout means wrong logic */
+	if (i == adev->usec_timeout)
+		return -ETIME;
+
+	return RREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90) == 0x1 ? 0 : -EIO;
+}
+
+static int smu_v11_0_send_msg(struct smu_context *smu, uint16_t msg)
+{
+	struct amdgpu_device *adev = smu->adev;
+	int ret = 0, index = 0;
+
+	index = smu_msg_get_index(smu, msg);
+	if (index < 0)
+		return index;
+
+	smu_v11_0_wait_for_response(smu);
+
+	WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0);
+
+	smu_v11_0_send_msg_without_waiting(smu, (uint16_t)index);
+
+	ret = smu_v11_0_wait_for_response(smu);
+
+	if (ret)
+		pr_err("Failed to send message 0x%x, response 0x%x\n", index,
+		       ret);
+
+	return ret;
+
+}
+
+static int
+smu_v11_0_send_msg_with_param(struct smu_context *smu, uint16_t msg,
+			      uint32_t param)
+{
+
+	struct amdgpu_device *adev = smu->adev;
+	int ret = 0, index = 0;
+
+	index = smu_msg_get_index(smu, msg);
+	if (index < 0)
+		return index;
+
+	ret = smu_v11_0_wait_for_response(smu);
+	if (ret)
+		pr_err("Failed to send message 0x%x, response 0x%x, param 0x%x\n",
+		       index, ret, param);
+
+	WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_90, 0);
+
+	WREG32_SOC15(MP1, 0, mmMP1_SMN_C2PMSG_82, param);
+
+	smu_v11_0_send_msg_without_waiting(smu, (uint16_t)index);
+
+	ret = smu_v11_0_wait_for_response(smu);
+	if (ret)
+		pr_err("Failed to send message 0x%x, response 0x%x param 0x%x\n",
+		       index, ret, param);
+
+	return ret;
+}
+
+static int smu_v11_0_init_microcode(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	const char *chip_name;
+	char fw_name[30];
+	int err = 0;
+	const struct smc_firmware_header_v1_0 *hdr;
+	const struct common_firmware_header *header;
+	struct amdgpu_firmware_info *ucode = NULL;
+
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+		chip_name = "vega20";
+		break;
+	case CHIP_ARCTURUS:
+		chip_name = "arcturus";
+		break;
+	default:
+		BUG();
+	}
+
+	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_smc.bin", chip_name);
+
+	err = request_firmware(&adev->pm.fw, fw_name, adev->dev);
+	if (err)
+		goto out;
+	err = amdgpu_ucode_validate(adev->pm.fw);
+	if (err)
+		goto out;
+
+	hdr = (const struct smc_firmware_header_v1_0 *) adev->pm.fw->data;
+	amdgpu_ucode_print_smc_hdr(&hdr->header);
+	adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
+
+	if (adev->firmware.load_type == AMDGPU_FW_LOAD_PSP) {
+		ucode = &adev->firmware.ucode[AMDGPU_UCODE_ID_SMC];
+		ucode->ucode_id = AMDGPU_UCODE_ID_SMC;
+		ucode->fw = adev->pm.fw;
+		header = (const struct common_firmware_header *)ucode->fw->data;
+		adev->firmware.fw_size +=
+			ALIGN(le32_to_cpu(header->ucode_size_bytes), PAGE_SIZE);
+	}
+
+out:
+	if (err) {
+		DRM_ERROR("smu_v11_0: Failed to load firmware \"%s\"\n",
+			  fw_name);
+		release_firmware(adev->pm.fw);
+		adev->pm.fw = NULL;
+	}
+	return err;
+}
+
+static int smu_v11_0_load_microcode(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	const uint32_t *src;
+	const struct smc_firmware_header_v1_0 *hdr;
+	uint32_t addr_start = MP1_SRAM;
+	uint32_t i;
+	uint32_t mp1_fw_flags;
+
+	hdr = (const struct smc_firmware_header_v1_0 *) adev->pm.fw->data;
+	src = (const uint32_t *)(adev->pm.fw->data +
+		le32_to_cpu(hdr->header.ucode_array_offset_bytes));
+
+	for (i = 1; i < MP1_SMC_SIZE/4 - 1; i++) {
+		if (amdgpu_emu_mode == 1 && i % 500 == 0)
+			msleep(1);
+		WREG32_PCIE(addr_start, src[i]);
+		addr_start += 4;
+	}
+
+	WREG32_PCIE(MP1_Public | (smnMP1_PUB_CTRL & 0xffffffff),
+		1 & MP1_SMN_PUB_CTRL__RESET_MASK);
+	WREG32_PCIE(MP1_Public | (smnMP1_PUB_CTRL & 0xffffffff),
+		1 & ~MP1_SMN_PUB_CTRL__RESET_MASK);
+
+	for (i = 0; i < adev->usec_timeout; i++) {
+		mp1_fw_flags = RREG32_PCIE(MP1_Public |
+			(smnMP1_FIRMWARE_FLAGS & 0xffffffff));
+		if ((mp1_fw_flags & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) >>
+			MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED__SHIFT)
+			break;
+		udelay(1);
+	}
+
+	if (i == adev->usec_timeout)
+		return -ETIME;
+
+	return 0;
+}
+
+static int smu_v11_0_check_fw_status(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	uint32_t mp1_fw_flags;
+
+	mp1_fw_flags = RREG32_PCIE(MP1_Public |
+				   (smnMP1_FIRMWARE_FLAGS & 0xffffffff));
+
+	if ((mp1_fw_flags & MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED_MASK) >>
+	    MP1_FIRMWARE_FLAGS__INTERRUPTS_ENABLED__SHIFT)
+		return 0;
+
+	return -EIO;
+}
+
+static int smu_v11_0_check_fw_version(struct smu_context *smu)
+{
+	uint32_t if_version = 0xff, smu_version = 0xff;
+	uint16_t smu_major;
+	uint8_t smu_minor, smu_debug;
+	int ret = 0;
+
+	ret = smu_get_smc_version(smu, &if_version, &smu_version);
+	if (ret)
+		return ret;
+
+	smu_major = (smu_version >> 16) & 0xffff;
+	smu_minor = (smu_version >> 8) & 0xff;
+	smu_debug = (smu_version >> 0) & 0xff;
+
+	pr_info("SMU Driver IF Version = 0x%08x, SMU FW Version = 0x%08x (%d.%d.%d)\n",
+		if_version, smu_version, smu_major, smu_minor, smu_debug);
+
+	if (if_version != smu->smc_if_version) {
+		pr_err("SMU driver if version not matched\n");
+		ret = -EINVAL;
+	}
+
+	return ret;
+}
+
+static int smu_v11_0_set_pptable_v2_0(struct smu_context *smu, void **table, uint32_t *size)
+{
+	struct amdgpu_device *adev = smu->adev;
+	uint32_t ppt_offset_bytes;
+	const struct smc_firmware_header_v2_0 *v2;
+
+	v2 = (const struct smc_firmware_header_v2_0 *) adev->pm.fw->data;
+
+	ppt_offset_bytes = le32_to_cpu(v2->ppt_offset_bytes);
+	*size = le32_to_cpu(v2->ppt_size_bytes);
+	*table = (uint8_t *)v2 + ppt_offset_bytes;
+
+	return 0;
+}
+
+static int smu_v11_0_set_pptable_v2_1(struct smu_context *smu, void **table, uint32_t *size, uint32_t pptable_id)
+{
+	struct amdgpu_device *adev = smu->adev;
+	const struct smc_firmware_header_v2_1 *v2_1;
+	struct smc_soft_pptable_entry *entries;
+	uint32_t pptable_count = 0;
+	int i = 0;
+
+	v2_1 = (const struct smc_firmware_header_v2_1 *) adev->pm.fw->data;
+	entries = (struct smc_soft_pptable_entry *)
+		((uint8_t *)v2_1 + le32_to_cpu(v2_1->pptable_entry_offset));
+	pptable_count = le32_to_cpu(v2_1->pptable_count);
+	for (i = 0; i < pptable_count; i++) {
+		if (le32_to_cpu(entries[i].id) == pptable_id) {
+			*table = ((uint8_t *)v2_1 + le32_to_cpu(entries[i].ppt_offset_bytes));
+			*size = le32_to_cpu(entries[i].ppt_size_bytes);
+			break;
+		}
+	}
+
+	if (i == pptable_count)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int smu_v11_0_setup_pptable(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	const struct smc_firmware_header_v1_0 *hdr;
+	int ret, index;
+	uint32_t size;
+	uint8_t frev, crev;
+	void *table;
+	uint16_t version_major, version_minor;
+
+	hdr = (const struct smc_firmware_header_v1_0 *) adev->pm.fw->data;
+	version_major = le16_to_cpu(hdr->header.header_version_major);
+	version_minor = le16_to_cpu(hdr->header.header_version_minor);
+	if (version_major == 2 && (smu->smu_table.boot_values.pp_table_id > 0 || amdgpu_emu_mode)) {
+		switch (version_minor) {
+		case 0:
+			ret = smu_v11_0_set_pptable_v2_0(smu, &table, &size);
+			break;
+		case 1:
+			if (amdgpu_emu_mode)
+				smu->smu_table.boot_values.pp_table_id = 2300;
+			ret = smu_v11_0_set_pptable_v2_1(smu, &table, &size,
+							 smu->smu_table.boot_values.pp_table_id);
+			break;
+		default:
+			ret = -EINVAL;
+			break;
+		}
+		if (ret)
+			return ret;
+
+	} else {
+		index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+						    powerplayinfo);
+
+		ret = smu_get_atom_data_table(smu, index, (uint16_t *)&size, &frev, &crev,
+					      (uint8_t **)&table);
+		if (ret)
+			return ret;
+	}
+
+	if (!smu->smu_table.power_play_table)
+		smu->smu_table.power_play_table = table;
+	if (!smu->smu_table.power_play_table_size)
+		smu->smu_table.power_play_table_size = size;
+
+	return 0;
+}
+
+static int smu_v11_0_init_dpm_context(struct smu_context *smu)
+{
+	struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
+
+	if (smu_dpm->dpm_context || smu_dpm->dpm_context_size != 0)
+		return -EINVAL;
+
+	return smu_alloc_dpm_context(smu);
+}
+
+static int smu_v11_0_fini_dpm_context(struct smu_context *smu)
+{
+	struct smu_dpm_context *smu_dpm = &smu->smu_dpm;
+
+	if (!smu_dpm->dpm_context || smu_dpm->dpm_context_size == 0)
+		return -EINVAL;
+
+	kfree(smu_dpm->dpm_context);
+	kfree(smu_dpm->golden_dpm_context);
+	kfree(smu_dpm->dpm_current_power_state);
+	kfree(smu_dpm->dpm_request_power_state);
+	smu_dpm->dpm_context = NULL;
+	smu_dpm->golden_dpm_context = NULL;
+	smu_dpm->dpm_context_size = 0;
+	smu_dpm->dpm_current_power_state = NULL;
+	smu_dpm->dpm_request_power_state = NULL;
+
+	return 0;
+}
+
+static int smu_v11_0_init_smc_tables(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	struct smu_table_context *smu_table = &smu->smu_table;
+	struct smu_table *tables = NULL;
+	int ret = 0;
+
+	if (smu_table->tables || smu_table->table_count != 0)
+		return -EINVAL;
+
+	tables = kcalloc(TABLE_COUNT, sizeof(struct smu_table), GFP_KERNEL);
+	if (!tables)
+		return -ENOMEM;
+
+	smu_table->tables = tables;
+	smu_table->table_count = TABLE_COUNT;
+
+	SMU_TABLE_INIT(tables, TABLE_PPTABLE, sizeof(PPTable_t),
+		       PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM);
+	SMU_TABLE_INIT(tables, TABLE_SMU_METRICS, sizeof(SmuMetrics_t),
+		       PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM);
+	ret = smu_v11_0_init_dpm_context(smu);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int smu_v11_0_fini_smc_tables(struct smu_context *smu)
+{
+	struct smu_table_context *smu_table = &smu->smu_table;
+	int ret = 0;
+
+	if (!smu_table->tables || smu_table->table_count == 0)
+		return -EINVAL;
+
+	kfree(smu_table->tables);
+	smu_table->tables = NULL;
+	smu_table->table_count = 0;
+
+	ret = smu_v11_0_fini_dpm_context(smu);
+	if (ret)
+		return ret;
+	return 0;
+}
+
+static int smu_v11_0_init_power(struct smu_context *smu)
+{
+	struct smu_power_context *smu_power = &smu->smu_power;
+
+	if (!smu->pm_enabled)
+		return 0;
+	if (smu_power->power_context || smu_power->power_context_size != 0)
+		return -EINVAL;
+
+	smu_power->power_context = kzalloc(sizeof(struct smu_11_0_dpm_context),
+					   GFP_KERNEL);
+	if (!smu_power->power_context)
+		return -ENOMEM;
+	smu_power->power_context_size = sizeof(struct smu_11_0_dpm_context);
+
+	smu->metrics_time = 0;
+	smu->metrics_table = kzalloc(sizeof(SmuMetrics_t), GFP_KERNEL);
+	if (!smu->metrics_table) {
+		kfree(smu_power->power_context);
+		return -ENOMEM;
+	}
+
+	return 0;
+}
+
+static int smu_v11_0_fini_power(struct smu_context *smu)
+{
+	struct smu_power_context *smu_power = &smu->smu_power;
+
+	if (!smu->pm_enabled)
+		return 0;
+	if (!smu_power->power_context || smu_power->power_context_size == 0)
+		return -EINVAL;
+
+	kfree(smu->metrics_table);
+	kfree(smu_power->power_context);
+	smu->metrics_table = NULL;
+	smu_power->power_context = NULL;
+	smu_power->power_context_size = 0;
+
+	return 0;
+}
+
+int smu_v11_0_get_vbios_bootup_values(struct smu_context *smu)
+{
+	int ret, index;
+	uint16_t size;
+	uint8_t frev, crev;
+	struct atom_common_table_header *header;
+	struct atom_firmware_info_v3_3 *v_3_3;
+	struct atom_firmware_info_v3_1 *v_3_1;
+	struct amdgpu_device *adev = smu->adev;
+
+	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
+					    firmwareinfo);
+
+	ret = smu_get_atom_data_table(smu, index, &size, &frev, &crev,
+				      (uint8_t **)&header);
+	if (ret)
+		return ret;
+
+	if (header->format_revision != 3) {
+		pr_err("unknown atom_firmware_info version! for smu11\n");
+		return -EINVAL;
+	}
+
+	switch (header->content_revision) {
+	case 0:
+	case 1:
+	case 2:
+		v_3_1 = (struct atom_firmware_info_v3_1 *)header;
+		smu->smu_table.boot_values.revision = v_3_1->firmware_revision;
+		smu->smu_table.boot_values.gfxclk = v_3_1->bootup_sclk_in10khz;
+		smu->smu_table.boot_values.uclk = v_3_1->bootup_mclk_in10khz;
+		smu->smu_table.boot_values.socclk = 0;
+		if (adev->asic_type != CHIP_ARCTURUS)
+			smu->smu_table.boot_values.dcefclk = 0;
+		smu->smu_table.boot_values.vddc = v_3_1->bootup_vddc_mv;
+		smu->smu_table.boot_values.vddci = v_3_1->bootup_vddci_mv;
+		smu->smu_table.boot_values.mvddc = v_3_1->bootup_mvddc_mv;
+		smu->smu_table.boot_values.vdd_gfx = v_3_1->bootup_vddgfx_mv;
+		smu->smu_table.boot_values.cooling_id = v_3_1->coolingsolution_id;
+		smu->smu_table.boot_values.pp_table_id = 0;
+		break;
+	case 3:
+	default:
+		v_3_3 = (struct atom_firmware_info_v3_3 *)header;
+		smu->smu_table.boot_values.revision = v_3_3->firmware_revision;
+		smu->smu_table.boot_values.gfxclk = v_3_3->bootup_sclk_in10khz;
+		smu->smu_table.boot_values.uclk = v_3_3->bootup_mclk_in10khz;
+		smu->smu_table.boot_values.socclk = 0;
+		if (adev->asic_type != CHIP_ARCTURUS)
+			smu->smu_table.boot_values.dcefclk = 0;
+		smu->smu_table.boot_values.vddc = v_3_3->bootup_vddc_mv;
+		smu->smu_table.boot_values.vddci = v_3_3->bootup_vddci_mv;
+		smu->smu_table.boot_values.mvddc = v_3_3->bootup_mvddc_mv;
+		smu->smu_table.boot_values.vdd_gfx = v_3_3->bootup_vddgfx_mv;
+		smu->smu_table.boot_values.cooling_id = v_3_3->coolingsolution_id;
+		smu->smu_table.boot_values.pp_table_id = v_3_3->pplib_pptable_id;
+	}
+
+	return 0;
+}
+
+static int smu_v11_0_get_clk_info_from_vbios(struct smu_context *smu)
+{
+	int ret, index;
+	struct amdgpu_device *adev = smu->adev;
+	struct atom_get_smu_clock_info_parameters_v3_1 input = {0};
+	struct atom_get_smu_clock_info_output_parameters_v3_1 *output;
+
+	input.clk_id = SMU11_SYSPLL0_SOCCLK_ID;
+	input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ;
+	index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1,
+					    getsmuclockinfo);
+
+	ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index,
+					(uint32_t *)&input);
+	if (ret)
+		return -EINVAL;
+
+	output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input;
+	smu->smu_table.boot_values.socclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000;
+
+	if (adev->asic_type != CHIP_ARCTURUS) {
+		memset(&input, 0, sizeof(input));
+		input.clk_id = SMU11_SYSPLL0_DCEFCLK_ID;
+		input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ;
+		index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1,
+						    getsmuclockinfo);
+
+		ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index,
+						(uint32_t *)&input);
+		if (ret)
+			return -EINVAL;
+
+		output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input;
+		smu->smu_table.boot_values.dcefclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000;
+
+		memset(&input, 0, sizeof(input));
+		input.clk_id = SMU11_SYSPLL0_ECLK_ID;
+		input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ;
+		index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1,
+						    getsmuclockinfo);
+
+		ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index,
+						(uint32_t *)&input);
+		if (ret)
+			return -EINVAL;
+
+		output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input;
+		smu->smu_table.boot_values.eclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000;
+
+		memset(&input, 0, sizeof(input));
+		input.clk_id = SMU11_SYSPLL0_VCLK_ID;
+		input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ;
+		index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1,
+						    getsmuclockinfo);
+
+		ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index,
+						(uint32_t *)&input);
+		if (ret)
+			return -EINVAL;
+
+		output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input;
+		smu->smu_table.boot_values.vclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000;
+
+		memset(&input, 0, sizeof(input));
+		input.clk_id = SMU11_SYSPLL0_DCLK_ID;
+		input.command = GET_SMU_CLOCK_INFO_V3_1_GET_CLOCK_FREQ;
+		index = get_index_into_master_table(atom_master_list_of_command_functions_v2_1,
+						    getsmuclockinfo);
+
+		ret = amdgpu_atom_execute_table(adev->mode_info.atom_context, index,
+						(uint32_t *)&input);
+		if (ret)
+			return -EINVAL;
+
+		output = (struct atom_get_smu_clock_info_output_parameters_v3_1 *)&input;
+		smu->smu_table.boot_values.dclk = le32_to_cpu(output->atom_smu_outputclkfreq.smu_clock_freq_hz) / 10000;
+	}
+
+	return 0;
+}
+
+static int smu_v11_0_notify_memory_pool_location(struct smu_context *smu)
+{
+	struct smu_table_context *smu_table = &smu->smu_table;
+	struct smu_table *memory_pool = &smu_table->memory_pool;
+	int ret = 0;
+	uint64_t address;
+	uint32_t address_low, address_high;
+
+	if (memory_pool->size == 0 || memory_pool->cpu_addr == NULL)
+		return ret;
+
+	address = (uintptr_t)memory_pool->cpu_addr;
+	address_high = (uint32_t)upper_32_bits(address);
+	address_low  = (uint32_t)lower_32_bits(address);
+
+	ret = smu_send_smc_msg_with_param(smu,
+					  SMU_MSG_SetSystemVirtualDramAddrHigh,
+					  address_high);
+	if (ret)
+		return ret;
+	ret = smu_send_smc_msg_with_param(smu,
+					  SMU_MSG_SetSystemVirtualDramAddrLow,
+					  address_low);
+	if (ret)
+		return ret;
+
+	address = memory_pool->mc_address;
+	address_high = (uint32_t)upper_32_bits(address);
+	address_low  = (uint32_t)lower_32_bits(address);
+
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DramLogSetDramAddrHigh,
+					  address_high);
+	if (ret)
+		return ret;
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DramLogSetDramAddrLow,
+					  address_low);
+	if (ret)
+		return ret;
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DramLogSetDramSize,
+					  (uint32_t)memory_pool->size);
+	if (ret)
+		return ret;
+
+	return ret;
+}
+
+static int smu_v11_0_check_pptable(struct smu_context *smu)
+{
+	int ret;
+
+	ret = smu_check_powerplay_table(smu);
+	return ret;
+}
+
+static int smu_v11_0_parse_pptable(struct smu_context *smu)
+{
+	int ret;
+
+	struct smu_table_context *table_context = &smu->smu_table;
+
+	if (table_context->driver_pptable)
+		return -EINVAL;
+
+	table_context->driver_pptable = kzalloc(sizeof(PPTable_t), GFP_KERNEL);
+
+	if (!table_context->driver_pptable)
+		return -ENOMEM;
+
+	ret = smu_store_powerplay_table(smu);
+	if (ret)
+		return -EINVAL;
+
+	if (amdgpu_emu_mode != 1)
+		ret = smu_append_powerplay_table(smu);
+
+	return ret;
+}
+
+static int smu_v11_0_populate_smc_pptable(struct smu_context *smu)
+{
+	int ret;
+
+	ret = smu_set_default_dpm_table(smu);
+
+	return ret;
+}
+
+static int smu_v11_0_write_pptable(struct smu_context *smu)
+{
+	struct smu_table_context *table_context = &smu->smu_table;
+	int ret = 0;
+
+	ret = smu_update_table(smu, TABLE_PPTABLE, table_context->driver_pptable, true);
+
+	return ret;
+}
+
+static int smu_v11_0_set_deep_sleep_dcefclk(struct smu_context *smu, uint32_t clk)
+{
+	int ret;
+
+	ret = smu_send_smc_msg_with_param(smu,
+					  SMU_MSG_SetMinDeepSleepDcefclk, clk);
+	if (ret)
+		pr_err("SMU11 attempt to set divider for DCEFCLK Failed!");
+
+	return ret;
+}
+
+static int smu_v11_0_set_min_dcef_deep_sleep(struct smu_context *smu)
+{
+	struct smu_table_context *table_context = &smu->smu_table;
+
+	if (!smu->pm_enabled)
+		return 0;
+	if (!table_context)
+		return -EINVAL;
+
+	return smu_set_deep_sleep_dcefclk(smu,
+					  table_context->boot_values.dcefclk / 100);
+}
+
+static int smu_v11_0_set_tool_table_location(struct smu_context *smu)
+{
+	int ret = 0;
+	struct smu_table *tool_table = &smu->smu_table.tables[TABLE_PMSTATUSLOG];
+
+	if (tool_table->mc_address) {
+		ret = smu_send_smc_msg_with_param(smu,
+				SMU_MSG_SetToolsDramAddrHigh,
+				upper_32_bits(tool_table->mc_address));
+		if (!ret)
+			ret = smu_send_smc_msg_with_param(smu,
+				SMU_MSG_SetToolsDramAddrLow,
+				lower_32_bits(tool_table->mc_address));
+	}
+
+	return ret;
+}
+
+static int smu_v11_0_init_display(struct smu_context *smu)
+{
+	int ret = 0;
+
+	if (!smu->pm_enabled)
+		return ret;
+
+	if (smu->adev->asic_type != CHIP_ARCTURUS)
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_NumOfDisplays, 0);
+
+	return ret;
+}
+
+static int smu_v11_0_update_feature_enable_state(struct smu_context *smu, uint32_t feature_id, bool enabled)
+{
+	uint32_t feature_low = 0, feature_high = 0;
+	int ret = 0;
+
+	if (!smu->pm_enabled)
+		return ret;
+	if (feature_id >= 0 && feature_id < 31)
+		feature_low = (1 << feature_id);
+	else if (feature_id > 31 && feature_id < 63)
+		feature_high = (1 << feature_id);
+	else
+		return -EINVAL;
+
+	if (enabled) {
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnableSmuFeaturesLow,
+						  feature_low);
+		if (ret)
+			return ret;
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_EnableSmuFeaturesHigh,
+						  feature_high);
+		if (ret)
+			return ret;
+
+	} else {
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DisableSmuFeaturesLow,
+						  feature_low);
+		if (ret)
+			return ret;
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_DisableSmuFeaturesHigh,
+						  feature_high);
+		if (ret)
+			return ret;
+
+	}
+
+	return ret;
+}
+
+static int smu_v11_0_set_allowed_mask(struct smu_context *smu)
+{
+	struct smu_feature *feature = &smu->smu_feature;
+	int ret = 0;
+	uint32_t feature_mask[2];
+
+	mutex_lock(&feature->mutex);
+	if (bitmap_empty(feature->allowed, SMU_FEATURE_MAX) || feature->feature_num < 64)
+		goto failed;
+
+	bitmap_copy((unsigned long *)feature_mask, feature->allowed, 64);
+
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetAllowedFeaturesMaskHigh,
+					  feature_mask[1]);
+	if (ret) {
+		printk("jack--> send set_allowed_high Failed!\n ");
+		goto failed;
+	} else {
+		printk("jack--> send set_allowed_high  Done!\n ");
+	}
+
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetAllowedFeaturesMaskLow,
+					  feature_mask[0]);
+	if (ret) {
+		printk("jack--> send set_allowed_low Failed!\n ");
+		goto failed;
+	} else {
+		printk("jack--> send set_allowed_low  Done!\n ");
+	}
+
+failed:
+	mutex_unlock(&feature->mutex);
+	return ret;
+}
+
+static int smu_v11_0_get_enabled_mask(struct smu_context *smu,
+				      uint32_t *feature_mask, uint32_t num)
+{
+	uint32_t feature_mask_high = 0, feature_mask_low = 0;
+	int ret = 0;
+
+	if (!feature_mask || num < 2)
+		return -EINVAL;
+
+	ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesHigh);
+	if (ret)
+		return ret;
+	ret = smu_read_smc_arg(smu, &feature_mask_high);
+	if (ret)
+		return ret;
+
+	ret = smu_send_smc_msg(smu, SMU_MSG_GetEnabledSmuFeaturesLow);
+	if (ret)
+		return ret;
+	ret = smu_read_smc_arg(smu, &feature_mask_low);
+	if (ret)
+		return ret;
+
+	feature_mask[0] = feature_mask_low;
+	feature_mask[1] = feature_mask_high;
+
+	return ret;
+}
+
+static bool smu_v11_0_is_dpm_running(struct smu_context *smu)
+{
+	int ret = 0;
+	uint32_t feature_mask[2];
+	unsigned long feature_enabled;
+	ret = smu_v11_0_get_enabled_mask(smu, feature_mask, 2);
+	feature_enabled = (unsigned long)((uint64_t)feature_mask[0] |
+			   ((uint64_t)feature_mask[1] << 32));
+	return !!(feature_enabled & SMC_DPM_FEATURE);
+}
+
+static int smu_v11_0_system_features_control(struct smu_context *smu,
+					     bool en)
+{
+	struct smu_feature *feature = &smu->smu_feature;
+	uint32_t feature_mask[2];
+	int ret = 0;
+
+	if (smu->pm_enabled) {
+		ret = smu_send_smc_msg(smu, (en ? SMU_MSG_EnableAllSmuFeatures :
+					     SMU_MSG_DisableAllSmuFeatures));
+		if (ret) {
+			printk("jack--> send enable_all_features Failed!\n ");
+			return ret;
+		} else {
+			printk("jack--> send enable_all_features Done!\n ");
+		}
+	}
+
+	ret = smu_feature_get_enabled_mask(smu, feature_mask, 2);
+	if (ret)
+		return ret;
+
+	bitmap_copy(feature->enabled, (unsigned long *)&feature_mask,
+		    feature->feature_num);
+	bitmap_copy(feature->supported, (unsigned long *)&feature_mask,
+		    feature->feature_num);
+
+	return ret;
+}
+
+static int smu_v11_0_notify_display_change(struct smu_context *smu)
+{
+	int ret = 0;
+
+	if (!smu->pm_enabled)
+		return ret;
+	if (smu_feature_is_enabled(smu, FEATURE_DPM_UCLK_BIT))
+	    ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetUclkFastSwitch, 1);
+
+	return ret;
+}
+
+static int
+smu_v11_0_get_max_sustainable_clock(struct smu_context *smu, uint32_t *clock,
+				    PPCLK_e clock_select)
+{
+	int ret = 0;
+
+	if (!smu->pm_enabled)
+		return ret;
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetDcModeMaxDpmFreq,
+					  clock_select << 16);
+	if (ret) {
+		pr_err("[GetMaxSustainableClock] Failed to get max DC clock from SMC!");
+		return ret;
+	}
+
+	ret = smu_read_smc_arg(smu, clock);
+	if (ret)
+		return ret;
+
+	if (*clock != 0)
+		return 0;
+
+	/* if DC limit is zero, return AC limit */
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetMaxDpmFreq,
+					  clock_select << 16);
+	if (ret) {
+		pr_err("[GetMaxSustainableClock] failed to get max AC clock from SMC!");
+		return ret;
+	}
+
+	ret = smu_read_smc_arg(smu, clock);
+
+	return ret;
+}
+
+static int smu_v11_0_init_max_sustainable_clocks(struct smu_context *smu)
+{
+	struct smu_11_0_max_sustainable_clocks *max_sustainable_clocks;
+	struct amdgpu_device *adev = smu->adev;
+	int ret = 0;
+
+	max_sustainable_clocks = kzalloc(sizeof(struct smu_11_0_max_sustainable_clocks),
+					 GFP_KERNEL);
+	smu->smu_table.max_sustainable_clocks = (void *)max_sustainable_clocks;
+
+	max_sustainable_clocks->uclock = smu->smu_table.boot_values.uclk / 100;
+	max_sustainable_clocks->soc_clock = smu->smu_table.boot_values.socclk / 100;
+	if (adev->asic_type != CHIP_ARCTURUS) {
+		max_sustainable_clocks->dcef_clock = smu->smu_table.boot_values.dcefclk / 100;
+		max_sustainable_clocks->display_clock = 0xFFFFFFFF;
+		max_sustainable_clocks->phy_clock = 0xFFFFFFFF;
+		max_sustainable_clocks->pixel_clock = 0xFFFFFFFF;
+	}
+
+	if (smu_feature_is_enabled(smu, FEATURE_DPM_UCLK_BIT)) {
+		ret = smu_v11_0_get_max_sustainable_clock(smu,
+							  &(max_sustainable_clocks->uclock),
+							  PPCLK_UCLK);
+		if (ret) {
+			pr_err("[%s] failed to get max UCLK from SMC!",
+			       __func__);
+			return ret;
+		}
+	}
+
+	if (smu_feature_is_enabled(smu, FEATURE_DPM_SOCCLK_BIT)) {
+		ret = smu_v11_0_get_max_sustainable_clock(smu,
+							  &(max_sustainable_clocks->soc_clock),
+							  PPCLK_SOCCLK);
+		if (ret) {
+			pr_err("[%s] failed to get max SOCCLK from SMC!",
+			       __func__);
+			return ret;
+		}
+	}
+
+	if (max_sustainable_clocks->soc_clock < max_sustainable_clocks->uclock)
+		max_sustainable_clocks->uclock = max_sustainable_clocks->soc_clock;
+
+	return 0;
+}
+
+static int smu_v11_0_get_current_clk_freq(struct smu_context *smu, uint32_t clk_id, uint32_t *value)
+{
+	static SmuMetrics_t metrics = {0};
+	int ret = 0;
+
+	if (!value)
+		return -EINVAL;
+
+	ret = smu_update_table(smu, TABLE_SMU_METRICS, (void *)&metrics, false);
+	if (ret)
+		return ret;
+
+	*value = metrics.CurrClock[clk_id];
+
+	return ret;
+}
+
+static int smu_v11_0_get_thermal_range(struct smu_context *smu,
+				struct PP_TemperatureRange *range)
+{
+	PPTable_t *pptable = smu->smu_table.driver_pptable;
+	memcpy(range, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
+
+	range->max = pptable->TedgeLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	range->edge_emergency_max = (pptable->TedgeLimit + CTF_OFFSET_EDGE) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	range->hotspot_crit_max = pptable->ThotspotLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	range->hotspot_emergency_max = (pptable->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+
+	return 0;
+}
+
+static int smu_v11_0_set_thermal_range(struct smu_context *smu,
+			struct PP_TemperatureRange *range)
+{
+	struct amdgpu_device *adev = smu->adev;
+	int low = SMU11_THERMAL_MINIMUM_ALERT_TEMP *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	int high = SMU11_THERMAL_MAXIMUM_ALERT_TEMP *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	uint32_t val;
+
+	if (low < range->min)
+		low = range->min;
+	if (high > range->max)
+		high = range->max;
+
+	if (low > high)
+		return -EINVAL;
+
+	val = RREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_CTRL);
+	val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, MAX_IH_CREDIT, 5);
+	val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, THERM_IH_HW_ENA, 1);
+	val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, DIG_THERM_INTH, (high / PP_TEMPERATURE_UNITS_PER_CENTIGRADES));
+	val = REG_SET_FIELD(val, THM_THERMAL_INT_CTRL, DIG_THERM_INTL, (low / PP_TEMPERATURE_UNITS_PER_CENTIGRADES));
+	val = val & (~THM_THERMAL_INT_CTRL__THERM_TRIGGER_MASK_MASK);
+
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_CTRL, val);
+
+	return 0;
+}
+
+static int smu_v11_0_enable_thermal_alert(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+	uint32_t val = 0;
+
+	val |= (1 << THM_THERMAL_INT_ENA__THERM_INTH_CLR__SHIFT);
+	val |= (1 << THM_THERMAL_INT_ENA__THERM_INTL_CLR__SHIFT);
+	val |= (1 << THM_THERMAL_INT_ENA__THERM_TRIGGER_CLR__SHIFT);
+
+	WREG32_SOC15(THM, 0, mmTHM_THERMAL_INT_ENA, val);
+
+	return 0;
+}
+
+static int smu_v11_0_set_thermal_fan_table(struct smu_context *smu)
+{
+	int ret;
+	struct smu_table_context *table_context = &smu->smu_table;
+	PPTable_t *pptable = table_context->driver_pptable;
+
+	ret = smu_send_smc_msg_with_param(smu, SMU_MSG_SetFanTemperatureTarget,
+			(uint32_t)pptable->FanTargetTemperature);
+
+	return ret;
+}
+
+static int smu_v11_0_start_thermal_control(struct smu_context *smu)
+{
+	int ret = 0;
+	struct PP_TemperatureRange range = {
+		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MAX};
+	struct amdgpu_device *adev = smu->adev;
+
+	if (!smu->pm_enabled)
+		return ret;
+	smu_v11_0_get_thermal_range(smu, &range);
+
+	if (smu->smu_table.thermal_controller_type) {
+		ret = smu_v11_0_set_thermal_range(smu, &range);
+		if (ret)
+			return ret;
+
+		ret = smu_v11_0_enable_thermal_alert(smu);
+		if (ret)
+			return ret;
+		if (adev->asic_type != CHIP_ARCTURUS) {
+			ret = smu_v11_0_set_thermal_fan_table(smu);
+			if (ret)
+				return ret;
+		}
+	}
+
+	adev->pm.dpm.thermal.min_temp = range.min;
+	adev->pm.dpm.thermal.max_temp = range.max;
+	adev->pm.dpm.thermal.max_edge_emergency_temp = range.edge_emergency_max;
+	adev->pm.dpm.thermal.min_hotspot_temp = range.hotspot_min;
+	adev->pm.dpm.thermal.max_hotspot_crit_temp = range.hotspot_crit_max;
+	adev->pm.dpm.thermal.max_hotspot_emergency_temp = range.hotspot_emergency_max;
+	adev->pm.dpm.thermal.min_mem_temp = range.mem_min;
+
+	return ret;
+}
+
+static int smu_v11_0_get_metrics_table(struct smu_context *smu,
+		SmuMetrics_t *metrics_table)
+{
+	int ret = 0;
+
+	if (!smu->metrics_time || time_after(jiffies, smu->metrics_time + HZ / 1000)) {
+		ret = smu_update_table(smu, TABLE_SMU_METRICS,
+				(void *)metrics_table, false);
+		if (ret) {
+			pr_info("Failed to export SMU metrics table!\n");
+			return ret;
+		}
+		memcpy(smu->metrics_table, metrics_table, sizeof(SmuMetrics_t));
+		smu->metrics_time = jiffies;
+	} else
+		memcpy(metrics_table, smu->metrics_table, sizeof(SmuMetrics_t));
+
+	return ret;
+}
+
+static int smu_v11_0_get_current_activity_percent(struct smu_context *smu,
+						  enum amd_pp_sensors sensor,
+						  uint32_t *value)
+{
+	int ret = 0;
+	SmuMetrics_t metrics;
+
+	if (!value)
+		return -EINVAL;
+
+	ret = smu_v11_0_get_metrics_table(smu, &metrics);
+	if (ret)
+		return ret;
+
+	switch (sensor) {
+	case AMDGPU_PP_SENSOR_GPU_LOAD:
+		*value = metrics.AverageGfxActivity;
+		break;
+	case AMDGPU_PP_SENSOR_MEM_LOAD:
+		*value = metrics.AverageUclkActivity;
+		break;
+	default:
+		pr_err("Invalid sensor for retrieving clock activity\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int smu_v11_0_thermal_get_temperature(struct smu_context *smu,
+					     enum amd_pp_sensors sensor,
+					     uint32_t *value)
+{
+	struct amdgpu_device *adev = smu->adev;
+	SmuMetrics_t metrics;
+	uint32_t temp = 0;
+	int ret = 0;
+
+	if (!value)
+		return -EINVAL;
+
+	ret = smu_v11_0_get_metrics_table(smu, &metrics);
+	if (ret)
+		return ret;
+
+	switch (sensor) {
+	case AMDGPU_PP_SENSOR_HOTSPOT_TEMP:
+		temp = RREG32_SOC15(THM, 0, mmCG_MULT_THERMAL_STATUS);
+		temp = (temp & CG_MULT_THERMAL_STATUS__CTF_TEMP_MASK) >>
+				CG_MULT_THERMAL_STATUS__CTF_TEMP__SHIFT;
+
+		temp = temp & 0x1ff;
+		temp *= SMU11_TEMPERATURE_UNITS_PER_CENTIGRADES;
+
+		*value = temp;
+		break;
+	case AMDGPU_PP_SENSOR_EDGE_TEMP:
+		*value = metrics.TemperatureEdge *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		break;
+	case AMDGPU_PP_SENSOR_MEM_TEMP:
+		*value = metrics.TemperatureHBM *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		break;
+	default:
+		pr_err("Invalid sensor for retrieving temp\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int smu_v11_0_get_gpu_power(struct smu_context *smu, uint32_t *value)
+{
+	int ret = 0;
+	SmuMetrics_t metrics;
+
+	if (!value)
+		return -EINVAL;
+
+	ret = smu_v11_0_get_metrics_table(smu, &metrics);
+	if (ret)
+		return ret;
+
+	*value = metrics.CurrSocketPower << 8;
+
+	return 0;
+}
+
+static int smu_v11_0_get_clock_ranges(struct smu_context *smu,
+				      uint32_t *clock,
+				      PPCLK_e clock_select,
+				      bool max)
+{
+	int ret;
+	*clock = 0;
+	if (max) {
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetMaxDpmFreq,
+					    (clock_select << 16));
+		if (ret) {
+			pr_err("[GetClockRanges] Failed to get max clock from SMC!\n");
+			return ret;
+		}
+		smu_read_smc_arg(smu, clock);
+	} else {
+		ret = smu_send_smc_msg_with_param(smu, SMU_MSG_GetMinDpmFreq,
+					    (clock_select << 16));
+		if (ret) {
+			pr_err("[GetClockRanges] Failed to get min clock from SMC!\n");
+			return ret;
+		}
+		smu_read_smc_arg(smu, clock);
+	}
+
+	return 0;
+}
+
+static uint32_t smu_v11_0_dpm_get_sclk(struct smu_context *smu, bool low)
+{
+	uint32_t gfx_clk;
+	int ret;
+
+	if (!smu_feature_is_enabled(smu, FEATURE_DPM_GFXCLK_BIT)) {
+		pr_err("[GetSclks]: gfxclk dpm not enabled!\n");
+		return -EPERM;
+	}
+
+	if (low) {
+		ret = smu_v11_0_get_clock_ranges(smu, &gfx_clk, PPCLK_GFXCLK, false);
+		if (ret) {
+			pr_err("[GetSclks]: fail to get min PPCLK_GFXCLK\n");
+			return ret;
+		}
+	} else {
+		ret = smu_v11_0_get_clock_ranges(smu, &gfx_clk, PPCLK_GFXCLK, true);
+		if (ret) {
+			pr_err("[GetSclks]: fail to get max PPCLK_GFXCLK\n");
+			return ret;
+		}
+	}
+
+	return (gfx_clk * 100);
+}
+
+static uint32_t smu_v11_0_dpm_get_mclk(struct smu_context *smu, bool low)
+{
+	uint32_t mem_clk;
+	int ret;
+
+	if (!smu_feature_is_enabled(smu, FEATURE_DPM_UCLK_BIT)) {
+		pr_err("[GetMclks]: memclk dpm not enabled!\n");
+		return -EPERM;
+	}
+
+	if (low) {
+		ret = smu_v11_0_get_clock_ranges(smu, &mem_clk, PPCLK_UCLK, false);
+		if (ret) {
+			pr_err("[GetMclks]: fail to get min PPCLK_UCLK\n");
+			return ret;
+		}
+	} else {
+		ret = smu_v11_0_get_clock_ranges(smu, &mem_clk, PPCLK_GFXCLK, true);
+		if (ret) {
+			pr_err("[GetMclks]: fail to get max PPCLK_UCLK\n");
+			return ret;
+		}
+	}
+
+	return (mem_clk * 100);
+}
+
+static const struct smu_funcs smu_v11_0_funcs = {
+	.init_microcode = smu_v11_0_init_microcode,
+	.load_microcode = smu_v11_0_load_microcode,
+	.check_fw_status = smu_v11_0_check_fw_status,
+	.check_fw_version = smu_v11_0_check_fw_version,
+	.send_smc_msg = smu_v11_0_send_msg,
+	.send_smc_msg_with_param = smu_v11_0_send_msg_with_param,
+	.read_smc_arg = smu_v11_0_read_arg,
+	.setup_pptable = smu_v11_0_setup_pptable,
+	.init_smc_tables = smu_v11_0_init_smc_tables,
+	.fini_smc_tables = smu_v11_0_fini_smc_tables,
+	.init_power = smu_v11_0_init_power,
+	.fini_power = smu_v11_0_fini_power,
+	.get_vbios_bootup_values = smu_v11_0_get_vbios_bootup_values,
+	.get_clk_info_from_vbios = smu_v11_0_get_clk_info_from_vbios,
+	.notify_memory_pool_location = smu_v11_0_notify_memory_pool_location,
+	.check_pptable = smu_v11_0_check_pptable,
+	.parse_pptable = smu_v11_0_parse_pptable,
+	.populate_smc_pptable = smu_v11_0_populate_smc_pptable,
+	.write_pptable = smu_v11_0_write_pptable,
+	.set_min_dcef_deep_sleep = smu_v11_0_set_min_dcef_deep_sleep,
+	.set_tool_table_location = smu_v11_0_set_tool_table_location,
+	.init_display = smu_v11_0_init_display,
+	.set_allowed_mask = smu_v11_0_set_allowed_mask,
+	.get_enabled_mask = smu_v11_0_get_enabled_mask,
+	.is_dpm_running = smu_v11_0_is_dpm_running,
+	.system_features_control = smu_v11_0_system_features_control,
+	.update_feature_enable_state = smu_v11_0_update_feature_enable_state,
+	.notify_display_change = smu_v11_0_notify_display_change,
+	.get_current_clk_freq = smu_v11_0_get_current_clk_freq,
+	.init_max_sustainable_clocks = smu_v11_0_init_max_sustainable_clocks,
+	.start_thermal_control = smu_v11_0_start_thermal_control,
+};
+
+void smu_v11_0_set_smu_funcs(struct smu_context *smu)
+{
+	struct amdgpu_device *adev = smu->adev;
+
+	smu->funcs = &smu_v11_0_funcs;
+	switch (adev->asic_type) {
+	case CHIP_VEGA20:
+		vega20_set_ppt_funcs(smu);
+		break;
+	case CHIP_ARCTURUS:
+		arcturus_set_ppt_funcs(smu);
+		break;
+	default:
+		pr_warn("Unknown asic for smu11\n");
+	}
+}
diff --git a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c
index f655ebd..adca84a 100644
--- a/drivers/gpu/drm/amd/powerplay/vega20_ppt.c
+++ b/drivers/gpu/drm/amd/powerplay/vega20_ppt.c
@@ -3135,6 +3135,28 @@ static int vega20_get_thermal_temperature_range(struct smu_context *smu,
 	return 0;
 }
 
+static int vega20_update_pcie_parameters(struct smu_context *smu,
+				     uint32_t pcie_gen_cap,
+				     uint32_t pcie_width_cap)
+{
+	PPTable_t *pptable = smu->smu_table.driver_pptable;
+	int ret, i;
+	uint32_t smu_pcie_arg;
+
+	for (i = 0; i < NUM_LINK_LEVELS; i++) {
+		smu_pcie_arg = (i << 16) |
+			((pptable->PcieGenSpeed[i] <= pcie_gen_cap) ? (pptable->PcieGenSpeed[i] << 8) :
+				(pcie_gen_cap << 8)) | ((pptable->PcieLaneCount[i] <= pcie_width_cap) ?
+					pptable->PcieLaneCount[i] : pcie_width_cap);
+		ret = smu_send_smc_msg_with_param(smu,
+					  SMU_MSG_OverridePcieParameters,
+					  smu_pcie_arg);
+	}
+
+	return ret;
+}
+
+
 static const struct pptable_funcs vega20_ppt_funcs = {
 	.tables_init = vega20_tables_init,
 	.alloc_dpm_context = vega20_allocate_dpm_context,
@@ -3177,7 +3199,8 @@ static const struct pptable_funcs vega20_ppt_funcs = {
 	.get_fan_speed_percent = vega20_get_fan_speed_percent,
 	.get_fan_speed_rpm = vega20_get_fan_speed_rpm,
 	.set_watermarks_table = vega20_set_watermarks_table,
-	.get_thermal_temperature_range = vega20_get_thermal_temperature_range
+	.get_thermal_temperature_range = vega20_get_thermal_temperature_range,
+	.update_pcie_parameters = vega20_update_pcie_parameters
 };
 
 void vega20_set_ppt_funcs(struct smu_context *smu)
-- 
2.7.4



More information about the amd-gfx mailing list