cpu/amd: Add CC6 support

This patch adds CC6 power save support to the AMD Family 15h
support code.  As CC6 is a complex power saving state that
relies heavily on CPU, northbridge, and southbridge cooperation,
this patch alters significant amounts of code throughout the
tree simultaneously.

Allowing the CPU to enter CC6 allows the second level of turbo
boost to be reached, and also provides significant power savings
when the system is idle due to the complete core shutdown.

Change-Id: I44ce157cda97fb85f3e8f3d7262d4712b5410670
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/11979
Tested-by: build bot (Jenkins)
Tested-by: Raptor Engineering Automated Test Stand <noreply@raptorengineeringinc.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>

20 files changed, 544 insertions, 201 deletions

diff --git a/src/arch/x86/acpigen.c b/src/arch/x86/acpigen.c
index cac2f8c9bb..c1a09f261d 100644
--- a/src/arch/x86/acpigen.c
+++ b/src/arch/x86/acpigen.c
@@ -1,6 +1,7 @@
 /*
  * This file is part of the coreboot project.
  *
+ * Copyright (C) 2015 Timothy Pearson <tpearson@raptorengineeringinc.com>, Raptor Engineering
  * Copyright (C) 2009 Rudolf Marek <r.marek@assembler.cz>
  *
  * This program is free software; you can redistribute it and/or modify
@@ -17,11 +18,11 @@
 #define ACPIGEN_LENSTACK_SIZE 10
 
 /*
- * If you need to change this, change acpigen_write_f and
+ * If you need to change this, change acpigen_write_len_f and
  * acpigen_pop_len
  */
 
-#define ACPIGEN_MAXLEN 0xfff
+#define ACPIGEN_MAXLEN 0xfffff
 
 #include <string.h>
 #include <arch/acpigen.h>
@@ -39,6 +40,7 @@ void acpigen_write_len_f(void)
 	len_stack[ltop++] = gencurrent;
 	acpigen_emit_byte(0);
 	acpigen_emit_byte(0);
+	acpigen_emit_byte(0);
 }
 
 void acpigen_pop_len(void)
@@ -48,9 +50,10 @@ void acpigen_pop_len(void)
 	char *p = len_stack[--ltop];
 	len = gencurrent - p;
 	ASSERT(len <= ACPIGEN_MAXLEN)
-	/* generate store length for 0xfff max */
-	p[0] = (0x40 | (len & 0xf));
+	/* generate store length for 0xfffff max */
+	p[0] = (0x80 | (len & 0xf));
 	p[1] = (len >> 4 & 0xff);
+	p[2] = (len >> 12 & 0xff);
 
 }
 
@@ -479,6 +482,21 @@ void acpigen_write_CST_package(acpi_cstate_t *cstate, int nentries)
 	acpigen_pop_len();
 }
 
+void acpigen_write_CSD_package(u32 domain, u32 numprocs, CSD_coord coordtype, u32 index)
+{
+	acpigen_write_name("_CSD");
+	acpigen_write_package(1);
+	acpigen_write_package(6);
+	acpigen_write_byte(6);	// 6 values
+	acpigen_write_byte(0);	// revision 0
+	acpigen_write_dword(domain);
+	acpigen_write_dword(coordtype);
+	acpigen_write_dword(numprocs);
+	acpigen_write_dword(index);
+	acpigen_pop_len();
+	acpigen_pop_len();
+}
+
 void acpigen_write_TSS_package(int entries, acpi_tstate_t *tstate_list)
 {
 /*
diff --git a/src/arch/x86/include/arch/acpigen.h b/src/arch/x86/include/arch/acpigen.h
index 038ec3ae3a..1ad021ada9 100644
--- a/src/arch/x86/include/arch/acpigen.h
+++ b/src/arch/x86/include/arch/acpigen.h
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2009 Rudolf Marek <r.marek@assembler.cz>
+ * Copyright (C) 2015 Timothy Pearson <tpearson@raptorengineeringinc.com>, Raptor Engineering
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -51,6 +52,8 @@ typedef enum { SW_ALL=0xfc, SW_ANY=0xfd, HW_ALL=0xfe } PSD_coord;
 void acpigen_write_PSD_package(u32 domain, u32 numprocs, PSD_coord coordtype);
 void acpigen_write_CST_package_entry(acpi_cstate_t *cstate);
 void acpigen_write_CST_package(acpi_cstate_t *entry, int nentries);
+typedef enum { CSD_HW_ALL=0xfe } CSD_coord;
+void acpigen_write_CSD_package(u32 domain, u32 numprocs, CSD_coord coordtype, u32 index);
 void acpigen_write_processor(u8 cpuindex, u32 pblock_addr, u8 pblock_len);
 void acpigen_write_TSS_package(int entries, acpi_tstate_t *tstate_list);
 void acpigen_write_TSD_package(u32 domain, u32 numprocs, PSD_coord coordtype);
diff --git a/src/cpu/amd/family_10h-family_15h/fidvid.c b/src/cpu/amd/family_10h-family_15h/fidvid.c
index 5eb7980514..3af18bd06b 100644
--- a/src/cpu/amd/family_10h-family_15h/fidvid.c
+++ b/src/cpu/amd/family_10h-family_15h/fidvid.c
@@ -165,87 +165,87 @@ static void applyBoostFIDOffset(device_t dev, uint32_t nodeid) {
 }
 
 static void enableNbPState1( device_t dev ) {
-  uint64_t cpuRev =  mctGetLogicalCPUID(0xFF);
-  if (cpuRev & AMD_FAM10_C3) {
-    u32 nbPState = (pci_read_config32(dev, 0x1F0) & NB_PSTATE_MASK);
-    if ( nbPState){
-      u32 nbVid1 = (pci_read_config32(dev, 0x1F4) & NB_VID1_MASK) >> NB_VID1_SHIFT;
-      u32 i;
-      for (i = nbPState; i < NM_PS_REG; i++) {
-         msr_t msr =  rdmsr(PS_REG_BASE + i);
-         if (msr.hi &  PS_EN_MASK ) {
-            msr.hi |= NB_DID_M_ON;
-            msr.lo &= NB_VID_MASK_OFF;
-	    msr.lo |= ( nbVid1 << NB_VID_POS);
-	    wrmsr(PS_REG_BASE + i, msr);
-	 }
-      }
-    }
-  }
+	uint64_t cpuRev =  mctGetLogicalCPUID(0xFF);
+	if (cpuRev & AMD_FAM10_C3) {
+		u32 nbPState = (pci_read_config32(dev, 0x1F0) & NB_PSTATE_MASK);
+		if ( nbPState){
+			u32 nbVid1 = (pci_read_config32(dev, 0x1F4) & NB_VID1_MASK) >> NB_VID1_SHIFT;
+			u32 i;
+			for (i = nbPState; i < NM_PS_REG; i++) {
+				msr_t msr =  rdmsr(PS_REG_BASE + i);
+				if (msr.hi &  PS_EN_MASK ) {
+				msr.hi |= NB_DID_M_ON;
+				msr.lo &= NB_VID_MASK_OFF;
+				msr.lo |= ( nbVid1 << NB_VID_POS);
+				wrmsr(PS_REG_BASE + i, msr);
+				}
+			}
+		}
+	}
 }
 
-static u8 setPStateMaxVal( device_t dev ) {
-      u8 i,maxpstate=0;
-      for (i = 0; i < NM_PS_REG; i++) {
-         msr_t msr =  rdmsr(PS_REG_BASE + i);
-         if (msr.hi & PS_IDD_VALUE_MASK) {
-	   msr.hi |= PS_EN_MASK ;
-	     wrmsr(PS_REG_BASE + i, msr);
-	 }
-         if (msr.hi & PS_EN_MASK) {
-	   maxpstate = i;
-	 }
-      }
-      //FIXME: CPTC2 and HTC_REG should get max per node, not per core ?
-      u32 reg = pci_read_config32(dev, CPTC2);
-      reg &= PS_MAX_VAL_MASK;
-      reg |= (maxpstate << PS_MAX_VAL_POS);
-      pci_write_config32(dev, CPTC2,reg);
-      return maxpstate;
+static u8 setPStateMaxVal(device_t dev) {
+	u8 i, maxpstate=0;
+	for (i = 0; i < NM_PS_REG; i++) {
+		msr_t msr = rdmsr(PS_REG_BASE + i);
+		if (msr.hi & PS_IDD_VALUE_MASK) {
+			msr.hi |= PS_EN_MASK ;
+			wrmsr(PS_REG_BASE + i, msr);
+		}
+		if (msr.hi & PS_EN_MASK) {
+			maxpstate = i;
+		}
+	}
+	//FIXME: CPTC2 and HTC_REG should get max per node, not per core ?
+	u32 reg = pci_read_config32(dev, CPTC2);
+	reg &= PS_MAX_VAL_MASK;
+	reg |= (maxpstate << PS_MAX_VAL_POS);
+	pci_write_config32(dev, CPTC2,reg);
+	return maxpstate;
 }
 
 static void dualPlaneOnly(  device_t dev ) {
-  // BKDG 2.4.2.7
-
-  uint64_t cpuRev =  mctGetLogicalCPUID(0xFF);
-  if ((mctGetProcessorPackageType() ==  AMD_PKGTYPE_AM3_2r2)
-      && (cpuRev & AMD_DR_Cx)) { // should be rev C or rev E but there's no constant for E
-    if ( (pci_read_config32(dev, 0x1FC) & DUAL_PLANE_ONLY_MASK)
-	 && (pci_read_config32(dev, 0xA0) & PVI_MODE) ){
-      if (cpuid_edx(0x80000007) & CPB_MASK) {
-          // revision E only, but E is apparently not supported yet, therefore untested
-         msr_t minPstate = rdmsr(0xC0010065);
-         wrmsr(0xC0010065, rdmsr(0xC0010068) );
-         wrmsr(0xC0010068,minPstate);
-      } else {
-	 msr_t msr;
-         msr.lo=0; msr.hi=0;
-         wrmsr(0xC0010064, rdmsr(0xC0010068) );
-         wrmsr(0xC0010068, msr );
-      }
-
-      //FIXME: CPTC2 and HTC_REG should get max per node, not per core ?
-      u8 maxpstate = setPStateMaxVal(dev);
-
-      u32 reg = pci_read_config32(dev, HTC_REG);
-      reg &= HTC_PS_LMT_MASK;
-      reg |= (maxpstate << PS_LIMIT_POS);
-      pci_write_config32(dev, HTC_REG,reg);
-
-    }
-  }
+	// BKDG 2.4.2.7
+
+	uint64_t cpuRev = mctGetLogicalCPUID(0xFF);
+	if ((mctGetProcessorPackageType() == AMD_PKGTYPE_AM3_2r2)
+		&& (cpuRev & (AMD_DR_Cx | AMD_DR_Ex))) {
+		if ((pci_read_config32(dev, 0x1FC) & DUAL_PLANE_ONLY_MASK)
+			&& (pci_read_config32(dev, 0xA0) & PVI_MODE)) {
+			if (cpuid_edx(0x80000007) & CPB_MASK) {
+				// revision E only, but E is apparently not supported yet, therefore untested
+				msr_t minPstate = rdmsr(0xC0010065);
+				wrmsr(0xC0010065, rdmsr(0xC0010068));
+				wrmsr(0xC0010068, minPstate);
+			} else {
+				msr_t msr;
+				msr.lo=0; msr.hi=0;
+				wrmsr(0xC0010064, rdmsr(0xC0010068) );
+				wrmsr(0xC0010068, msr);
+			}
+
+			//FIXME: CPTC2 and HTC_REG should get max per node, not per core ?
+			u8 maxpstate = setPStateMaxVal(dev);
+
+			u32 reg = pci_read_config32(dev, HTC_REG);
+			reg &= HTC_PS_LMT_MASK;
+			reg |= (maxpstate << PS_LIMIT_POS);
+			pci_write_config32(dev, HTC_REG,reg);
+		}
+	}
 }
 
 static int vidTo100uV(u8 vid)
-{// returns voltage corresponding to vid in tenths of mV, i.e. hundreds of uV
- // BKDG #31116 rev 3.48 2.4.1.6
-  int voltage;
-  if (vid >= 0x7c) {
-    voltage = 0;
-  } else {
-    voltage = (15500 - (125*vid));
-  }
-  return voltage;
+{
+	// returns voltage corresponding to vid in tenths of mV, i.e. hundreds of uV
+	// BKDG #31116 rev 3.48 2.4.1.6
+	int voltage;
+	if (vid >= 0x7c) {
+		voltage = 0;
+	} else {
+		voltage = (15500 - (125*vid));
+	}
+	return voltage;
 }
 
 static void setVSRamp(device_t dev) {
@@ -344,7 +344,7 @@ static void recalculateVsSlamTimeSettingOnCorePre(device_t dev)
 	}
 
 	/* Get AltVID */
-	dtemp = pci_read_config32(dev, 0xDC);
+	dtemp = pci_read_config32(dev, 0xdc);
 	bValue = (u8) (dtemp & BIT_MASK_7);
 
 	/* Use the VID with the lowest voltage (higher VID) */
@@ -508,15 +508,15 @@ static void config_nb_syn_ptr_adj(device_t dev, u32 cpuRev) {
            values (min latency) */
 	u32 nbPstate = pci_read_config32(dev,0x1f0) & NB_PSTATE_MASK;
         u8 nbSynPtrAdj;
-	if ((cpuRev & (AMD_DR_Bx|AMD_DA_Cx) )
-	    || ( (cpuRev & AMD_RB_C3) && (nbPstate!=0)))  {
-	  nbSynPtrAdj = 5;
+	if ((cpuRev & (AMD_DR_Bx | AMD_DA_Cx | AMD_FAM15_ALL) )
+		|| ((cpuRev & AMD_RB_C3) && (nbPstate != 0))) {
+		nbSynPtrAdj = 5;
 	} else {
-          nbSynPtrAdj = 6;
+		nbSynPtrAdj = 6;
 	}
 
-	u32 dword = pci_read_config32(dev, 0xDc);
-        dword &= ~ NB_SYN_PTR_ADJ_MASK;
+	u32 dword = pci_read_config32(dev, 0xdc);
+        dword &= ~NB_SYN_PTR_ADJ_MASK;
 	dword |= nbSynPtrAdj << NB_SYN_PTR_ADJ_POS;
         /* NbsynPtrAdj set to 5 or 6 per BKDG (needs reset) */
 	pci_write_config32(dev, 0xdc, dword);
@@ -548,7 +548,7 @@ static void config_acpi_pwr_state_ctrl_regs(device_t dev, u32 cpuRev, u8 procPkg
 			}
 		} else { // rev C or later
 			// same doubt as cache scrubbing: ok to check current state ?
-			dword = pci_read_config32(dev, 0xDC);
+			dword = pci_read_config32(dev, 0xdc);
 			u32 cacheFlushOnHalt = dword & (7 << 16);
 			if (!cacheFlushOnHalt) {
 				c1 = 0x80;
@@ -619,11 +619,11 @@ static void prep_fid_change(void)
 		printk(BIOS_DEBUG, "  F3x80: %08x\n", dword);
 		dword = pci_read_config32(dev, 0x84);
 		printk(BIOS_DEBUG, "  F3x84: %08x\n", dword);
-		dword = pci_read_config32(dev, 0xD4);
+		dword = pci_read_config32(dev, 0xd4);
 		printk(BIOS_DEBUG, "  F3xD4: %08x\n", dword);
-		dword = pci_read_config32(dev, 0xD8);
+		dword = pci_read_config32(dev, 0xd8);
 		printk(BIOS_DEBUG, "  F3xD8: %08x\n", dword);
-		dword = pci_read_config32(dev, 0xDC);
+		dword = pci_read_config32(dev, 0xdc);
 		printk(BIOS_DEBUG, "  F3xDC: %08x\n", dword);
 	}
 }
@@ -752,7 +752,7 @@ static void fixPsNbVidBeforeWR(u32 newNbVid, u32 coreid, u32 dev, u8 pviMode)
          * synchronization between cores and we don't think
          * PstatMaxVal is going to be 0 on cold reset anyway ?
 	 */
-        if ( ! (pci_read_config32(dev, 0xDC) & (~ PS_MAX_VAL_MASK)) ) {
+        if (!(pci_read_config32(dev, 0xdc) & (~PS_MAX_VAL_MASK))) {
   	   printk(BIOS_ERR,"F3xDC[PstateMaxVal] is zero. Northbridge voltage setting will fail. fixPsNbVidBeforeWR in fidvid.c needs fixing. See AMD # 31116 rev 3.48 BKDG 2.4.2.9.1 \n");
 	};
 
diff --git a/src/cpu/amd/family_10h-family_15h/init_cpus.c b/src/cpu/amd/family_10h-family_15h/init_cpus.c
index 85eb931ce9..c9dca765c4 100644
--- a/src/cpu/amd/family_10h-family_15h/init_cpus.c
+++ b/src/cpu/amd/family_10h-family_15h/init_cpus.c
@@ -26,6 +26,14 @@
 #include <northbridge/amd/amdfam10/raminit_amdmct.c>
 #include <reset.h>
 
+#if IS_ENABLED(CONFIG_SOUTHBRIDGE_AMD_SB700)
+#include <southbridge/amd/sb700/sb700.h>
+#endif
+
+#if IS_ENABLED(CONFIG_SOUTHBRIDGE_AMD_SB800)
+#include <southbridge/amd/sb800/sb800.h>
+#endif
+
 #if IS_ENABLED(CONFIG_SET_FIDVID)
 static void prep_fid_change(void);
 static void init_fidvid_stage2(u32 apicid, u32 nodeid);
@@ -870,6 +878,7 @@ void cpuSetAMDMSR(uint8_t node_id)
 	u8 i;
 	u32 platform;
 	uint64_t revision;
+	uint8_t enable_c_states;
 
 	printk(BIOS_DEBUG, "cpuSetAMDMSR ");
 
@@ -932,6 +941,44 @@ void cpuSetAMDMSR(uint8_t node_id)
 		wrmsr(FP_CFG, msr);
 	}
 
+#if IS_ENABLED(CONFIG_SOUTHBRIDGE_AMD_SB700) || IS_ENABLED(CONFIG_SOUTHBRIDGE_AMD_SB800)
+	uint8_t nvram;
+
+	if (revision & (AMD_DR_GT_D0 | AMD_FAM15_ALL)) {
+		/* Set up message triggered C1E */
+		msr = rdmsr(0xc0010055);
+		msr.lo &= ~0xffff;		/* IOMsgAddr = ACPI_PM_EVT_BLK */
+		msr.lo |= ACPI_PM_EVT_BLK & 0xffff;
+		msr.lo |= (0x1 << 29);		/* BmStsClrOnHltEn = 1 */
+		if (revision & AMD_DR_GT_D0) {
+			msr.lo &= ~(0x1 << 28);	/* C1eOnCmpHalt = 0 */
+			msr.lo &= ~(0x1 << 27);	/* SmiOnCmpHalt = 0 */
+		}
+		wrmsr(0xc0010055, msr);
+
+		msr = rdmsr(0xc0010015);
+		msr.lo |= (0x1 << 12);		/* HltXSpCycEn = 1 */
+		wrmsr(0xc0010015, msr);
+	}
+
+	if (revision & (AMD_DR_Ex | AMD_FAM15_ALL)) {
+		enable_c_states = 0;
+		if (IS_ENABLED(CONFIG_HAVE_ACPI_TABLES))
+			if (get_option(&nvram, "cpu_c_states") == CB_SUCCESS)
+				enable_c_states = !!nvram;
+
+		if (enable_c_states) {
+			/* Set up the C-state base address */
+			msr_t c_state_addr_msr;
+			c_state_addr_msr = rdmsr(0xc0010073);
+			c_state_addr_msr.lo = ACPI_CPU_P_LVL2;	/* CstateAddr = ACPI_CPU_P_LVL2 */
+			wrmsr(0xc0010073, c_state_addr_msr);
+		}
+	}
+#else
+	enable_c_states = 0;
+#endif
+
 	printk(BIOS_DEBUG, " done\n");
 }
 
@@ -946,6 +993,7 @@ static void cpuSetAMDPCI(u8 node)
 	u32 platform;
 	u32 val;
 	u8 offset;
+	uint32_t dword;
 	uint64_t revision;
 
 	printk(BIOS_DEBUG, "cpuSetAMDPCI %02d", node);
@@ -1004,6 +1052,39 @@ static void cpuSetAMDPCI(u8 node)
 	   if (revision & (AMD_DR_B2 | AMD_DR_B3))
 	   dctPhyDiag(); */
 
+	if (revision & (AMD_DR_GT_D0 | AMD_FAM15_ALL)) {
+		/* Set up message triggered C1E */
+		dword = pci_read_config32(NODE_PCI(node, 3), 0xd4);
+		dword &= ~(0x1 << 14);			/* CacheFlushImmOnAllHalt = !is_fam15h() */
+		dword |= (is_fam15h()?0:1) << 14;
+		pci_write_config32(NODE_PCI(node, 3), 0xd4, dword);
+
+		dword = pci_read_config32(NODE_PCI(node, 3), 0xdc);
+		dword |= 0x1 << 26;			/* IgnCpuPrbEn = 1 */
+		dword &= ~(0x7f << 19);			/* CacheFlushOnHaltTmr = 0x28 */
+		dword |= 0x28 << 19;
+		dword |= 0x7 << 16;			/* CacheFlushOnHaltCtl = 0x7 */
+		pci_write_config32(NODE_PCI(node, 3), 0xdc, dword);
+
+		dword = pci_read_config32(NODE_PCI(node, 3), 0xa0);
+		dword |= 0x1 << 10;			/* IdleExitEn = 1 */
+		pci_write_config32(NODE_PCI(node, 3), 0xa0, dword);
+
+		if (revision & AMD_DR_GT_D0) {
+			dword = pci_read_config32(NODE_PCI(node, 3), 0x188);
+			dword |= 0x1 << 4;			/* EnStpGntOnFlushMaskWakeup = 1 */
+			pci_write_config32(NODE_PCI(node, 3), 0x188, dword);
+		} else {
+			dword = pci_read_config32(NODE_PCI(node, 4), 0x128);
+			dword &= ~(0x1 << 31);			/* CstateMsgDis = 0 */
+			pci_write_config32(NODE_PCI(node, 4), 0x128, dword);
+		}
+
+		dword = pci_read_config32(NODE_PCI(node, 3), 0xd4);
+		dword |= 0x1 << 13;			/* MTC1eEn = 1 */
+		pci_write_config32(NODE_PCI(node, 3), 0xd4, dword);
+	}
+
 	printk(BIOS_DEBUG, " done\n");
 }
 
diff --git a/src/cpu/amd/family_10h-family_15h/powernow_acpi.c b/src/cpu/amd/family_10h-family_15h/powernow_acpi.c
index 0c53e86b4a..a7cc9dad5b 100644
--- a/src/cpu/amd/family_10h-family_15h/powernow_acpi.c
+++ b/src/cpu/amd/family_10h-family_15h/powernow_acpi.c
@@ -17,6 +17,7 @@
 
 #include <console/console.h>
 #include <stdint.h>
+#include <option.h>
 #include <cpu/x86/msr.h>
 #include <arch/acpigen.h>
 #include <cpu/amd/powernow.h>
@@ -30,21 +31,29 @@
 #include <northbridge/amd/amdmct/mct/mct.h>
 #include <northbridge/amd/amdmct/amddefs.h>
 
+static inline uint8_t is_fam15h(void)
+{
+	uint8_t fam15h = 0;
+	uint32_t family;
+
+	family = cpuid_eax(0x80000001);
+	family = ((family & 0xf00000) >> 16) | ((family & 0xf00) >> 8);
+
+	if (family >= 0x6f)
+		/* Family 15h or later */
+		fam15h = 1;
+
+	return fam15h;
+}
+
 static void write_pstates_for_core(u8 pstate_num, u16 *pstate_feq, u32 *pstate_power,
 				u32 *pstate_latency, u32 *pstate_control,
 				u32 *pstate_status, int coreID,
-				u32 pcontrol_blk, u8 plen, u8 onlyBSP,
 				uint8_t single_link)
 {
 	int i;
 	struct cpuid_result cpuid1;
 
-	if ((onlyBSP) && (coreID != 0)) {
-	    plen = 0;
-	    pcontrol_blk = 0;
-	}
-
-	acpigen_write_processor(coreID, pcontrol_blk, plen);
 	acpigen_write_empty_PCT();
 	acpigen_write_name("_PSS");
 
@@ -88,9 +97,62 @@ static void write_pstates_for_core(u8 pstate_num, u16 *pstate_feq, u32 *pstate_p
 		if (cpu)
 			acpigen_write_PSD_package(cpu->path.apic.apic_id, 1, SW_ANY);
 	}
+}
 
-	/* patch the whole Processor token length */
-	acpigen_pop_len();
+static void write_cstates_for_core(int coreID)
+{
+	/* Generate C state entries */
+	uint8_t cstate_count = 1;
+	acpi_cstate_t cstate;
+
+	if (is_fam15h()) {
+		cstate.ctype = 2;
+		cstate.latency = 100;
+		cstate.power = 0;
+		cstate.resource.space_id = ACPI_ADDRESS_SPACE_IO;
+		cstate.resource.bit_width = 8;
+		cstate.resource.bit_offset = 0;
+		cstate.resource.addrl = rdmsr(0xc0010073).lo + 1;
+		cstate.resource.addrh = 0;
+		cstate.resource.resv = 1;
+	} else {
+		cstate.ctype = 2;
+		cstate.latency = 75;
+		cstate.power = 0;
+		cstate.resource.space_id = ACPI_ADDRESS_SPACE_IO;
+		cstate.resource.bit_width = 8;
+		cstate.resource.bit_offset = 0;
+		cstate.resource.addrl = rdmsr(0xc0010073).lo;
+		cstate.resource.addrh = 0;
+		cstate.resource.resv = 1;
+	}
+
+	acpigen_write_CST_package(&cstate, cstate_count);
+
+	/* Find the local APIC ID for the specified core ID */
+	if (is_fam15h()) {
+		struct device* cpu;
+		int cpu_index = 0;
+		for (cpu = all_devices; cpu; cpu = cpu->next) {
+			if ((cpu->path.type != DEVICE_PATH_APIC) ||
+				(cpu->bus->dev->path.type != DEVICE_PATH_CPU_CLUSTER))
+				continue;
+			if (!cpu->enabled)
+				continue;
+			if (cpu_index == coreID)
+				break;
+			cpu_index++;
+		}
+
+		if (cpu) {
+			/* TODO
+			 * Detect dual core status and skip CSD generation if dual core is disabled
+			 */
+
+			/* Generate C state dependency entries */
+			acpigen_write_CSD_package((cpu->path.apic.apic_id >> 1) & 0x7f, 2, CSD_HW_ALL, 0);
+		}
+	}
 }
 
 /*
@@ -121,6 +183,15 @@ void amd_generate_powernow(u32 pcontrol_blk, u8 plen, u8 onlyBSP)
 	u8 index;
 	msr_t msr;
 
+	uint8_t nvram;
+	uint8_t enable_c_states;
+
+	enable_c_states = 0;
+#if IS_ENABLED(CONFIG_HAVE_ACPI_TABLES)
+	if (get_option(&nvram, "cpu_c_states") == CB_SUCCESS)
+		enable_c_states = !!nvram;
+#endif
+
 	/* Get the Processor Brand String using cpuid(0x8000000x) command x=2,3,4 */
 	cpuid1 = cpuid(0x80000002);
 	v = (u32 *) processor_brand;
@@ -196,6 +267,10 @@ void amd_generate_powernow(u32 pcontrol_blk, u8 plen, u8 onlyBSP)
 		return;
 	}
 
+	if (fam15h)
+		/* Set P_LVL2 P_BLK entry */
+		*(((uint8_t *)pcontrol_blk) + 0x04) = (rdmsr(0xc0010073).lo + 1) & 0xff;
+
 	uint8_t pviModeFlag;
 	uint8_t Pstate_max;
 	uint8_t cpufid;
@@ -314,18 +389,56 @@ void amd_generate_powernow(u32 pcontrol_blk, u8 plen, u8 onlyBSP)
 			    Pstate_latency[index]);
 	}
 
+	/* Enter processor block scope */
 	char pscope[] = "\\_PR";
-
 	acpigen_write_scope(pscope);
+
 	for (index = 0; index < total_core_count; index++) {
 		/* Determine if this is a single-link processor */
 		node_index = 0x18 + (index / cores_per_node);
 		dtemp = pci_read_config32(dev_find_slot(0, PCI_DEVFN(node_index, 0)), 0x80);
 		single_link = !!(((dtemp & 0xff00) >> 8) == 0);
 
+		/* Enter processor core scope */
+		uint8_t plen_cur = plen;
+		uint32_t pcontrol_blk_cur = pcontrol_blk;
+		if ((onlyBSP) && (index != 0)) {
+			plen_cur = 0;
+			pcontrol_blk_cur = 0;
+		}
+		acpigen_write_processor(index, pcontrol_blk_cur, plen_cur);
+
+		/* Write P-state status and dependency objects */
 		write_pstates_for_core(Pstate_num, Pstate_feq, Pstate_power,
 				Pstate_latency, Pstate_control, Pstate_status,
-				index, pcontrol_blk, plen, onlyBSP, single_link);
+				index, single_link);
+
+		/* Write C-state status and dependency objects */
+		if (fam15h && enable_c_states)
+			write_cstates_for_core(index);
+
+		/* Exit processor core scope */
+		acpigen_pop_len();
 	}
+
+	/* Exit processor block scope */
 	acpigen_pop_len();
 }
+
+void amd_powernow_update_fadt(acpi_fadt_t * fadt)
+{
+	if (is_fam15h()) {
+		fadt->p_lvl2_lat = 101;		/* NOTE: While the BKDG states this should
+						 * be set to 100, there is no way to meet
+						 * the other FADT requirements.  I suspect
+						 * there is an error in the BKDG for ACPI
+						 * 1.x support; disable all FADT-based C
+						 * states > 2... */
+		fadt->p_lvl3_lat = 1001;
+		fadt->flags |= 0x1 << 2;	/* FLAGS.PROC_C1 = 1 */
+		fadt->flags |= 0x1 << 3;	/* FLAGS.P_LVL2_UP = 1 */
+	} else {
+		fadt->cst_cnt = 0;
+	}
+	fadt->pstate_cnt = 0;
+}
diff --git a/src/include/cpu/amd/powernow.h b/src/include/cpu/amd/powernow.h
index 5f749d00b8..fc1e8635e8 100644
--- a/src/include/cpu/amd/powernow.h
+++ b/src/include/cpu/amd/powernow.h
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2009 Rudolf Marek <r.marek@assembler.cz>
+ * Copyright (C) 2015 Timothy Pearson <tpearson@raptorengineeringinc.com>, Raptor Engineering
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -17,5 +18,6 @@
 #define POWERNOW_H
 
 void amd_generate_powernow(u32 pcontrol_blk, u8 plen, u8 onlyBSP);
+void amd_powernow_update_fadt(acpi_fadt_t * fadt);
 
 #endif
diff --git a/src/mainboard/asus/kgpe-d16/cmos.default b/src/mainboard/asus/kgpe-d16/cmos.default
index b42541af58..c841aa9503 100644
--- a/src/mainboard/asus/kgpe-d16/cmos.default
+++ b/src/mainboard/asus/kgpe-d16/cmos.default
@@ -14,6 +14,7 @@ ecc_scrub_rate = 1.28us
 interleave_chip_selects = Enable
 interleave_nodes = Disable
 interleave_memory_channels = Enable
+cpu_c_states = Enable
 cpu_cc6_state = Enable
 ieee1394_controller = Enable
 power_on_after_fail = On
diff --git a/src/mainboard/asus/kgpe-d16/cmos.layout b/src/mainboard/asus/kgpe-d16/cmos.layout
index fb6ea738b4..40799d5c55 100644
--- a/src/mainboard/asus/kgpe-d16/cmos.layout
+++ b/src/mainboard/asus/kgpe-d16/cmos.layout
@@ -38,8 +38,9 @@ entries
 458          4       e       11       hypertransport_speed_limit
 462          2       e       12       minimum_memory_voltage
 464          1       e       2        compute_unit_siblings
-465          1       e       1        cpu_cc6_state
-466          1       r       0        allow_spd_nvram_cache_restore
+465          1       e       1        cpu_c_states
+466          1       e       1        cpu_cc6_state
+467          1       r       0        allow_spd_nvram_cache_restore
 477          1       e       1        ieee1394_controller
 728        256       h       0        user_data
 984         16       h       0        check_sum
diff --git a/src/northbridge/amd/amdfam10/link_control.c b/src/northbridge/amd/amdfam10/link_control.c
index 1091ef4f2e..7fa9f12b05 100644
--- a/src/northbridge/amd/amdfam10/link_control.c
+++ b/src/northbridge/amd/amdfam10/link_control.c
@@ -49,15 +49,94 @@ static inline uint8_t is_fam15h(void)
 
 static void nb_control_init(struct device *dev)
 {
+	uint8_t enable_c_states;
+	uint8_t enable_cc6;
 	uint32_t dword;
 
 	printk(BIOS_DEBUG, "NB: Function 4 Link Control.. ");
 
+	/* Configure L3 Power Control */
+	dword = pci_read_config32(dev, 0x1c4);
+	dword |= (0x1 << 8);			/* L3PwrSavEn = 1 */
+	pci_write_config32(dev, 0x1c4, dword);
+
 	if (is_fam15h()) {
+		/* Configure L3 Control 2 */
+		dword = pci_read_config32(dev, 0x1cc);
+		dword &= ~(0x7 << 6);			/* ImplRdProjDelayThresh = 0x2 */
+		dword |= (0x2 << 6);
+		pci_write_config32(dev, 0x1cc, dword);
+
+		/* Configure TDP Accumulator Divisor Control */
+		dword = pci_read_config32(dev, 0x104);
+		dword &= ~(0xfff << 2);			/* TdpAccDivRate = 0xc8 */
+		dword |= (0xc8 << 2);
+		dword &= ~0x3;				/* TdpAccDivVal = 0x1 */
+		dword |= 0x1;
+		pci_write_config32(dev, 0x104, dword);
+
+		/* Configure Sample and Residency Timers */
+		dword = pci_read_config32(dev, 0x110);
+		dword &= ~0xfff;			/* CSampleTimer = 0x1 */
+		dword |= 0x1;
+		pci_write_config32(dev, 0x110, dword);
+
+		/* Configure APM TDP Control */
+		dword = pci_read_config32(dev, 0x16c);
+		dword |= (0x1 << 4);			/* ApmTdpLimitIntEn = 1 */
+		pci_write_config32(dev, 0x16c, dword);
+
 		/* Enable APM */
 		dword = pci_read_config32(dev, 0x15c);
 		dword |= (0x1 << 7);			/* ApmMasterEn = 1 */
 		pci_write_config32(dev, 0x15c, dword);
+
+		enable_c_states = 0;
+		enable_cc6 = 0;
+#if IS_ENABLED(CONFIG_HAVE_ACPI_TABLES)
+		uint8_t nvram;
+
+		if (get_option(&nvram, "cpu_c_states") == CB_SUCCESS)
+			enable_c_states = !!nvram;
+
+		if (get_option(&nvram, "cpu_cc6_state") == CB_SUCCESS)
+			enable_cc6 = !!nvram;
+#endif
+
+		if (enable_c_states) {
+			/* Configure C-state Control 1 */
+			dword = pci_read_config32(dev, 0x118);
+			dword |= (0x1 << 24);		/* PwrGateEnCstAct1 = 1 */
+			dword &= ~(0x7 << 21);		/* ClkDivisorCstAct1 = 0x0 */
+			dword &= ~(0x3 << 18);		/* CacheFlushTmrSelCstAct1 = 0x1 */
+			dword |= (0x1 << 18);
+			dword |= (0x1 << 17);		/* CacheFlushEnCstAct1 = 1 */
+			dword |= (0x1 << 16);		/* CpuPrbEnCstAct1 = 1 */
+			dword &= ~(0x1 << 8);		/* PwrGateEnCstAct0 = 0 */
+			dword &= ~(0x7 << 5);		/* ClkDivisorCstAct0 = 0x0 */
+			dword &= ~(0x3 << 2);		/* CacheFlushTmrSelCstAct0 = 0x2 */
+			dword |= (0x2 << 2);
+			dword |= (0x1 << 1);		/* CacheFlushEnCstAct0 = 1 */
+			dword |= 0x1;			/* CpuPrbEnCstAct0 = 1 */
+			pci_write_config32(dev, 0x118, dword);
+
+			/* Configure C-state Control 2 */
+			dword = pci_read_config32(dev, 0x11c);
+			dword &= ~(0x1 << 8);		/* PwrGateEnCstAct2 = 0 */
+			dword &= ~(0x7 << 5);		/* ClkDivisorCstAct2 = 0x0 */
+			dword &= ~(0x3 << 2);		/* CacheFlushTmrSelCstAct0 = 0x0 */
+			dword &= ~(0x1 << 1);		/* CacheFlushEnCstAct0 = 0 */
+			dword &= ~(0x1);		/* CpuPrbEnCstAct0 = 0 */
+			pci_write_config32(dev, 0x11c, dword);
+
+			/* Configure C-state Policy Control 1 */
+			dword = pci_read_config32(dev, 0x128);
+			dword &= ~(0x7f << 5);		/* CacheFlushTmr = 0x28 */
+			dword |= (0x28 << 5);
+			dword &= ~0x1;			/* CoreCstateMode = !enable_cc6 */
+			dword |= ((enable_cc6)?0:1);
+			pci_write_config32(dev, 0x128, dword);
+		}
 	}
 
 	printk(BIOS_DEBUG, "done.\n");
@@ -83,4 +162,4 @@ static const struct pci_driver mcf4_driver_fam15 __pci_driver = {
 	.ops    = &mcf4_ops,
 	.vendor = PCI_VENDOR_ID_AMD,
 	.device = 0x1604,
-};
\ No newline at end of file
+};
diff --git a/src/northbridge/amd/amdfam10/northbridge.c b/src/northbridge/amd/amdfam10/northbridge.c
index 116789ce11..0c1971bbc4 100644
--- a/src/northbridge/amd/amdfam10/northbridge.c
+++ b/src/northbridge/amd/amdfam10/northbridge.c
@@ -766,53 +766,49 @@ static void amdfam10_domain_read_resources(device_t dev)
 			uint8_t num_nodes;
 
 			/* Find highest DRAM range (DramLimitAddr) */
+			num_nodes = 0;
 			max_node = 0;
 			max_range = -1;
 			interleaved = 0;
 			max_range_limit = 0;
-			for (range = 0; range < 8; range++) {
-				dword = f1_read_config32(0x40 + (range * 0x8));
-				if (!(dword & 0x3))
-					continue;
-
-				if ((dword >> 8) & 0x7)
-					interleaved = 1;
-
-				dword = f1_read_config32(0x44 + (range * 0x8));
-				dword2 = f1_read_config32(0x144 + (range * 0x8));
-				qword = ((((uint64_t)dword) >> 16) & 0xffff) << 24;
-				qword |= (((uint64_t)dword2) & 0xff) << 40;
-
-				if (qword > max_range_limit) {
-					max_range = range;
-					max_range_limit = qword;
-					max_node = dword & 0x7;
-				}
-			}
-
-			num_nodes = 0;
 			device_t node_dev;
 			for (node = 0; node < FX_DEVS; node++) {
 				node_dev = get_node_pci(node, 0);
 				/* Test for node presence */
-				if ((node_dev) && (pci_read_config32(node_dev, PCI_VENDOR_ID) != 0xffffffff))
-					num_nodes++;
+				if ((!node_dev) || (pci_read_config32(node_dev, PCI_VENDOR_ID) == 0xffffffff))
+					continue;
+
+				num_nodes++;
+				for (range = 0; range < 8; range++) {
+					dword = pci_read_config32(get_node_pci(node, 1), 0x40 + (range * 0x8));
+					if (!(dword & 0x3))
+						continue;
+
+					if ((dword >> 8) & 0x7)
+						interleaved = 1;
+
+					dword = pci_read_config32(get_node_pci(node, 1), 0x44 + (range * 0x8));
+					dword2 = pci_read_config32(get_node_pci(node, 1), 0x144 + (range * 0x8));
+					qword = 0xffffff;
+					qword |= ((((uint64_t)dword) >> 16) & 0xffff) << 24;
+					qword |= (((uint64_t)dword2) & 0xff) << 40;
+
+					if (qword > max_range_limit) {
+						max_range = range;
+						max_range_limit = qword;
+						max_node = dword & 0x7;
+					}
+				}
 			}
 
-			/* Calculate CC6 sotrage area size */
+			/* Calculate CC6 storage area size */
 			if (interleaved)
 				qword = (0x1000000 * num_nodes);
 			else
 				qword = 0x1000000;
 
 			/* Reserve the CC6 save segment */
-			reserved_ram_resource(dev, 8, max_range_limit >> 10, qword >> 10);
-
-			/* Set up the C-state base address */
-			msr_t c_state_addr_msr;
-			c_state_addr_msr = rdmsr(0xc0010073);
-			c_state_addr_msr.lo = 0xe0e0;		/* CstateAddr = 0xe0e0 */
-			wrmsr(0xc0010073, c_state_addr_msr);
+			reserved_ram_resource(dev, 8, (max_range_limit + 1) >> 10, qword >> 10);
 		}
 	}
 }
diff --git a/src/northbridge/amd/amdht/AsPsDefs.h b/src/northbridge/amd/amdht/AsPsDefs.h
index 559c67d990..d4e6a29f42 100644
--- a/src/northbridge/amd/amdht/AsPsDefs.h
+++ b/src/northbridge/amd/amdht/AsPsDefs.h
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2007-2008 Advanced Micro Devices, Inc.
+ * Copyright (C) 2015 Timothy Pearson <tpearson@raptorengineeringinc.com>, Raptor Engineering
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -250,7 +251,7 @@
 #define DUAL_PLANE_NB_VID_SHIFT 17/* for CPU rev <= C */
 
 
-#define NM_PS_REG 5			/* number of P-state MSR registers */
+#define NM_PS_REG (is_fam15h()?8:5)	/* number of P-state MSR registers */
 
 /* sFidVidInit.outFlags defines */
 #define PWR_CK_OK 0			/* System board check OK */
diff --git a/src/northbridge/amd/amdmct/amddefs.h b/src/northbridge/amd/amdmct/amddefs.h
index 42ad32b1e2..d2dfbccfe9 100644
--- a/src/northbridge/amd/amdmct/amddefs.h
+++ b/src/northbridge/amd/amdmct/amddefs.h
@@ -49,32 +49,34 @@
 /*
  * Groups - Create as many as you wish, from the above public values
  */
-#define	AMD_NPT_F2	(AMD_NPT_F2C | AMD_NPT_F2D | AMD_NPT_F2E | AMD_NPT_F2G | AMD_NPT_F2J | AMD_NPT_F2K)
-#define	AMD_NPT_F3	(AMD_NPT_F3L)
-#define	AMD_NPT_Fx	(AMD_NPT_F0 | AMD_NPT_F1 | AMD_NPT_F2 | AMD_NPT_F3)
-#define	AMD_NPT_Gx	(AMD_NPT_G0A | AMD_NPT_G1B)
-#define	AMD_NPT_ALL	(AMD_NPT_Fx | AMD_NPT_Gx)
-#define	AMD_FINEDELAY	(AMD_NPT_F0 | AMD_NPT_F1 | AMD_NPT_F2)
-#define	AMD_GT_F0	(AMD_NPT_ALL AND NOT AMD_NPT_F0)
-#define	AMD_DR_Ax	(AMD_DR_A0A + AMD_DR_A1B + AMD_DR_A2)
-#define	AMD_DR_Bx	(AMD_DR_B0 | AMD_DR_B1 | AMD_DR_B2 | AMD_DR_B3 | AMD_DR_BA)
-#define	AMD_DR_LT_B2	(AMD_DR_B0 | AMD_DR_B1 | AMD_DR_BA)
-#define	AMD_DR_LT_B3	(AMD_DR_B0 | AMD_DR_B1 | AMD_DR_B2 | AMD_DR_BA)
-#define	AMD_DR_GT_B0	(AMD_DR_ALL & ~(AMD_DR_B0))
-#define	AMD_DR_GT_Bx	(AMD_DR_ALL & ~(AMD_DR_Ax | AMD_DR_Bx))
-#define	AMD_DR_ALL	(AMD_DR_Bx)
-#define	AMD_FAM10_ALL	(AMD_DR_ALL | AMD_RB_C2 | AMD_HY_D0 | AMD_DA_C3 | AMD_DA_C2 | AMD_RB_C3 | AMD_HY_D1 | AMD_PH_E0)
-#define AMD_FAM10_LT_D  (AMD_FAM10_ALL & ~(AMD_HY_D0))
-#define	AMD_FAM10_GT_B0	(AMD_FAM10_ALL & ~(AMD_DR_B0))
-#define AMD_FAM10_REV_D	(AMD_HY_D0 | AMD_HY_D1)
-#define	AMD_DA_Cx       (AMD_DA_C2 | AMD_DA_C3)
-#define	AMD_DR_Cx       (AMD_RB_C2 | AMD_RB_C3 | AMD_DA_Cx)
-#define	AMD_FAM10_C3       (AMD_RB_C3 | AMD_DA_C3)
-#define	AMD_DR_Dx	(AMD_HY_D0 | AMD_HY_D1)
-#define	AMD_DRBH_Cx	(AMD_DR_Cx | AMD_HY_D0 )
-#define	AMD_DRBA23_RBC2	(AMD_DR_BA | AMD_DR_B2 | AMD_DR_B3 | AMD_RB_C2 )
+#define	AMD_NPT_F2		(AMD_NPT_F2C | AMD_NPT_F2D | AMD_NPT_F2E | AMD_NPT_F2G | AMD_NPT_F2J | AMD_NPT_F2K)
+#define	AMD_NPT_F3		(AMD_NPT_F3L)
+#define	AMD_NPT_Fx		(AMD_NPT_F0 | AMD_NPT_F1 | AMD_NPT_F2 | AMD_NPT_F3)
+#define	AMD_NPT_Gx		(AMD_NPT_G0A | AMD_NPT_G1B)
+#define	AMD_NPT_ALL		(AMD_NPT_Fx | AMD_NPT_Gx)
+#define	AMD_FINEDELAY		(AMD_NPT_F0 | AMD_NPT_F1 | AMD_NPT_F2)
+#define	AMD_GT_F0		(AMD_NPT_ALL AND NOT AMD_NPT_F0)
+#define	AMD_DR_Ax		(AMD_DR_A0A + AMD_DR_A1B + AMD_DR_A2)
+#define	AMD_DR_Bx		(AMD_DR_B0 | AMD_DR_B1 | AMD_DR_B2 | AMD_DR_B3 | AMD_DR_BA)
+#define	AMD_DR_Cx		(AMD_RB_C2 | AMD_RB_C3 | AMD_DA_Cx)
+#define	AMD_DR_Dx		(AMD_HY_D0 | AMD_HY_D1)
+#define	AMD_DR_Ex		(AMD_PH_E0)
+#define	AMD_DR_LT_B2		(AMD_DR_B0 | AMD_DR_B1 | AMD_DR_BA)
+#define	AMD_DR_LT_B3		(AMD_DR_B0 | AMD_DR_B1 | AMD_DR_B2 | AMD_DR_BA)
+#define	AMD_DR_GT_B0		(AMD_DR_ALL & ~(AMD_DR_B0))
+#define	AMD_DR_GT_Bx		(AMD_DR_ALL & ~(AMD_DR_Ax | AMD_DR_Bx))
+#define	AMD_DR_GT_D0		((AMD_DR_Dx & ~(AMD_HY_D0)) | AMD_DR_Ex)
+#define	AMD_DR_ALL		(AMD_DR_Bx)
+#define	AMD_FAM10_ALL		(AMD_DR_ALL | AMD_RB_C2 | AMD_HY_D0 | AMD_DA_C3 | AMD_DA_C2 | AMD_RB_C3 | AMD_HY_D1 | AMD_PH_E0)
+#define AMD_FAM10_LT_D  	(AMD_FAM10_ALL & ~(AMD_HY_D0))
+#define	AMD_FAM10_GT_B0		(AMD_FAM10_ALL & ~(AMD_DR_B0))
+#define AMD_FAM10_REV_D		(AMD_HY_D0 | AMD_HY_D1)
+#define	AMD_DA_Cx       	(AMD_DA_C2 | AMD_DA_C3)
+#define	AMD_FAM10_C3		(AMD_RB_C3 | AMD_DA_C3)
+#define	AMD_DRBH_Cx		(AMD_DR_Cx | AMD_HY_D0 )
+#define	AMD_DRBA23_RBC2		(AMD_DR_BA | AMD_DR_B2 | AMD_DR_B3 | AMD_RB_C2 )
 #define	AMD_DR_DAC2_OR_C3	(AMD_DA_C2 | AMD_DA_C3 | AMD_RB_C3)
-#define	AMD_FAM15_ALL	(AMD_OR_B2 | AMD_OR_C0)
+#define	AMD_FAM15_ALL		(AMD_OR_B2 | AMD_OR_C0)
 
 /*
  *  Public Platforms - USE THESE VERSIONS TO MAKE COMPARE WITH CPUPLATFORMTYPE RETURN VALUE
@@ -87,9 +89,9 @@
 #define	AMD_PTYPE_MC	0x020	/* Multi Core (>2) */
 #define	AMD_PTYPE_UMA	0x040	/* UMA required */
 
-	/*
-	 * Groups - Create as many as you wish, from the above public values
-	 */
+/*
+ * Groups - Create as many as you wish, from the above public values
+ */
 #define	AMD_PTYPE_ALL	0xFFFFFFFF	/* A mask for all */
 
 
@@ -98,11 +100,11 @@
  */
 #define HTPHY_LINKTYPE_HT3		0x00000001
 #define HTPHY_LINKTYPE_HT1		0x00000002
-#define HTPHY_LINKTYPE_COHERENT	0x00000004
+#define HTPHY_LINKTYPE_COHERENT		0x00000004
 #define HTPHY_LINKTYPE_NONCOHERENT	0x00000008
 #define HTPHY_LINKTYPE_CONNECTED	(HTPHY_LINKTYPE_COHERENT | HTPHY_LINKTYPE_NONCOHERENT)
 #define HTPHY_LINKTYPE_GANGED		0x00000010
-#define HTPHY_LINKTYPE_UNGANGED	0x00000020
+#define HTPHY_LINKTYPE_UNGANGED		0x00000020
 #define HTPHY_LINKTYPE_ALL		0x7FFFFFFF
 
 
@@ -110,7 +112,7 @@
  * CPU HT PHY REGISTERS, FIELDS, AND MASKS
  */
 #define HTPHY_OFFSET_MASK		0xE00001FF
-#define HTPHY_WRITE_CMD		0x40000000
+#define HTPHY_WRITE_CMD			0x40000000
 #define HTPHY_IS_COMPLETE_MASK		0x80000000
 #define HTPHY_DIRECT_MAP		0x20000000
 #define HTPHY_DIRECT_OFFSET_MASK	0xE000FFFF
@@ -158,4 +160,4 @@
 #define AMD_PKGTYPE_S1gX 2
 #define AMD_PKGTYPE_G34 3
 #define AMD_PKGTYPE_ASB2 4
-#define AMD_PKGTYPE_C32 5
+#define AMD_PKGTYPE_C32 5
\ No newline at end of file
diff --git a/src/northbridge/amd/amdmct/mct_ddr3/mct_d.c b/src/northbridge/amd/amdmct/mct_ddr3/mct_d.c
index 27fa7ae018..1d48070084 100644
--- a/src/northbridge/amd/amdmct/mct_ddr3/mct_d.c
+++ b/src/northbridge/amd/amdmct/mct_ddr3/mct_d.c
@@ -1203,6 +1203,7 @@ static void set_up_cc6_storage_fam15(struct MCTStatStruc *pMCTstat,
 	int8_t max_range;
 	uint8_t max_node;
 	uint64_t max_range_limit;
+	uint8_t byte;
 	uint32_t dword;
 	uint32_t dword2;
 	uint64_t qword;
@@ -1222,7 +1223,8 @@ static void set_up_cc6_storage_fam15(struct MCTStatStruc *pMCTstat,
 
 		dword = Get_NB32(pDCTstat->dev_map, 0x44 + (range * 0x8));
 		dword2 = Get_NB32(pDCTstat->dev_map, 0x144 + (range * 0x8));
-		qword = ((((uint64_t)dword) >> 16) & 0xffff) << 24;
+		qword = 0xffffff;
+		qword |= ((((uint64_t)dword) >> 16) & 0xffff) << 24;
 		qword |= (((uint64_t)dword2) & 0xff) << 40;
 
 		if (qword > max_range_limit) {
@@ -1232,26 +1234,35 @@ static void set_up_cc6_storage_fam15(struct MCTStatStruc *pMCTstat,
 		}
 	}
 
-	if (pDCTstat->Node_ID == max_node) {
-		if (max_range >= 0) {
-			if (interleaved)
-				/* Move upper limit down by 16M * the number of nodes */
-				max_range_limit -= (0x1000000 * num_nodes);
-			else
-				/* Move upper limit down by 16M */
-				max_range_limit -= 0x1000000;
-
-			/* Store modified range */
-			dword = Get_NB32(pDCTstat->dev_map, 0x44 + (range * 0x8));
-			dword &= ~(0xffff << 16);		/* DramLimit[39:24] = max_range_limit[39:24] */
-			dword |= (max_range_limit >> 24) & 0xffff;
-			Set_NB32(pDCTstat->dev_map, 0x44 + (range * 0x8), dword);
-
-			dword = Get_NB32(pDCTstat->dev_map, 0x144 + (range * 0x8));
-			dword &= ~(0xffff << 16);		/* DramLimit[47:40] = max_range_limit[47:40] */
-			dword |= (max_range_limit >> 40) & 0xff;
-			Set_NB32(pDCTstat->dev_map, 0x144 + (range * 0x8), dword);
-		}
+	if (max_range >= 0) {
+		if (interleaved)
+			/* Move upper limit down by 16M * the number of nodes */
+			max_range_limit -= (0x1000000 * num_nodes);
+		else
+			/* Move upper limit down by 16M */
+			max_range_limit -= 0x1000000;
+
+		/* Disable the range */
+		dword = Get_NB32(pDCTstat->dev_map, 0x40 + (max_range * 0x8));
+		byte = dword & 0x3;
+		dword &= ~(0x3);
+		Set_NB32(pDCTstat->dev_map, 0x40 + (max_range * 0x8), dword);
+
+		/* Store modified range */
+		dword = Get_NB32(pDCTstat->dev_map, 0x44 + (max_range * 0x8));
+		dword &= ~(0xffff << 16);		/* DramLimit[39:24] = max_range_limit[39:24] */
+		dword |= ((max_range_limit >> 24) & 0xffff) << 16;
+		Set_NB32(pDCTstat->dev_map, 0x44 + (max_range * 0x8), dword);
+
+		dword = Get_NB32(pDCTstat->dev_map, 0x144 + (max_range * 0x8));
+		dword &= ~0xff;			/* DramLimit[47:40] = max_range_limit[47:40] */
+		dword |= (max_range_limit >> 40) & 0xff;
+		Set_NB32(pDCTstat->dev_map, 0x144 + (max_range * 0x8), dword);
+
+		/* Reenable the range */
+		dword = Get_NB32(pDCTstat->dev_map, 0x40 + (max_range * 0x8));
+		dword |= byte;
+		Set_NB32(pDCTstat->dev_map, 0x40 + (max_range * 0x8), dword);
 	}
 
 	/* Determine save state destination node */
@@ -1538,8 +1549,8 @@ restartinit:
 					pDCTstat = pDCTstatA + Node;
 
 					if (pDCTstat->NodePresent) {
-						lock_dram_config(pMCTstat, pDCTstat);
 						set_cc6_save_enable(pMCTstat, pDCTstat, 1);
+						lock_dram_config(pMCTstat, pDCTstat);
 					}
 				}
 			}
@@ -5118,7 +5129,7 @@ static void mct_HTMemMapExt(struct MCTStatStruc *pMCTstat,
 		/* get base/limit from Node0 */
 		reg = 0x40 + (Node << 3);		/* Node0/Dram Base 0 */
 		val = Get_NB32(dev, reg);
-		Drambase = val >> ( 16 + 3);
+		Drambase = val >> (16 + 3);
 
 		reg = 0x44 + (Node << 3);		/* Node0/Dram Base 0 */
 		val = Get_NB32(dev, reg);
diff --git a/src/northbridge/amd/amdmct/mct_ddr3/mctecc_d.c b/src/northbridge/amd/amdmct/mct_ddr3/mctecc_d.c
index 0a51b1a228..6ab33a3b1e 100644
--- a/src/northbridge/amd/amdmct/mct_ddr3/mctecc_d.c
+++ b/src/northbridge/amd/amdmct/mct_ddr3/mctecc_d.c
@@ -155,6 +155,14 @@ u8 ECCInit_D(struct MCTStatStruc *pMCTstat, struct DCTStatStruc *pDCTstatA)
 			if (MemClrECC) {
 				MCTMemClrSync_D(pMCTstat, pDCTstatA);
 			}
+
+			if (pDCTstat->LogicalCPUID & (AMD_DR_GT_D0 | AMD_FAM15_ALL)) {
+				/* Set up message triggered C1E */
+				val = pci_read_config32(pDCTstat->dev_nbmisc, 0xd4);
+				val &= ~(0x1 << 15);			/* StutterScrubEn = DRAM scrub enabled */
+				val |= (mctGet_NVbits(NV_DramBKScrub)?1:0) << 15;
+				pci_write_config32(pDCTstat->dev_nbmisc, 0xd4, val);
+			}
 		}	/* if Node present */
 	}
 
diff --git a/src/southbridge/amd/sb700/early_setup.c b/src/southbridge/amd/sb700/early_setup.c
index 57fda78077..f109896ea3 100644
--- a/src/southbridge/amd/sb700/early_setup.c
+++ b/src/southbridge/amd/sb700/early_setup.c
@@ -268,10 +268,6 @@ void enable_fid_change_on_sb(u32 sbbusn, u32 sbdn)
 	byte &= ~(1<<6);
 	pmio_write(0x8d, byte);
 
-	byte = pmio_read(0x61);
-	byte &= ~0x04;
-	pmio_write(0x61, byte);
-
 	byte = pmio_read(0x42);
 	byte &= ~0x04;
 	pmio_write(0x42, byte);
@@ -583,6 +579,13 @@ static void sb700_devices_por_init(void)
 static void sb700_pmio_por_init(void)
 {
 	u8 byte;
+	uint8_t enable_c_states;
+
+	enable_c_states = 0;
+#if IS_ENABLED(CONFIG_HAVE_ACPI_TABLES)
+	if (get_option(&byte, "cpu_c_states") == CB_SUCCESS)
+		enable_c_states = !!byte;
+#endif
 
 	printk(BIOS_INFO, "sb700_pmio_por_init()\n");
 	/* K8KbRstEn, KB_RST# control for K8 system. */
@@ -644,6 +647,14 @@ static void sb700_pmio_por_init(void)
 	byte |= 1 << 0;
 	pmio_write(0xB2, byte);
 
+	/* Set up IOAPIC and BM_STS monitoring */
+	byte = pmio_read(0x61);
+	if (enable_c_states)
+		byte |= 0x4;
+	else
+		byte &= ~0x04;
+	pmio_write(0x61, byte);
+
 	/* NOTE: Enabling automatic C1e state switch caused failures when initializing processors */
 
 	/* Enable precision HPET clock and automatic C state switch */
diff --git a/src/southbridge/amd/sb700/fadt.c b/src/southbridge/amd/sb700/fadt.c
index 7a8a1a6762..2417cebaca 100644
--- a/src/southbridge/amd/sb700/fadt.c
+++ b/src/southbridge/amd/sb700/fadt.c
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2010 Advanced Micro Devices, Inc.
+ * Copyright (C) 2015 Raptor Engineering
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -22,6 +23,7 @@
 #include <arch/acpi.h>
 #include <arch/io.h>
 #include <device/device.h>
+#include <cpu/amd/powernow.h>
 #include "sb700.h"
 
 void acpi_create_fadt(acpi_fadt_t * fadt, acpi_facs_t * facs, void *dsdt)
@@ -152,5 +154,8 @@ void acpi_create_fadt(acpi_fadt_t * fadt, acpi_facs_t * facs, void *dsdt)
 	fadt->x_gpe1_blk.addrl = 0;
 	fadt->x_gpe1_blk.addrh = 0x0;
 
+	if (IS_ENABLED(CONFIG_CPU_AMD_MODEL_10XXX))
+		amd_powernow_update_fadt(fadt);
+
 	header->checksum = acpi_checksum((void *)fadt, sizeof(acpi_fadt_t));
 }
diff --git a/src/southbridge/amd/sb700/sb700.h b/src/southbridge/amd/sb700/sb700.h
index a610267561..f895811472 100644
--- a/src/southbridge/amd/sb700/sb700.h
+++ b/src/southbridge/amd/sb700/sb700.h
@@ -32,10 +32,11 @@
 
 #define ACPI_PM_EVT_BLK		(SB700_ACPI_IO_BASE + 0x00) /* 4 bytes */
 #define ACPI_PM1_CNT_BLK	(SB700_ACPI_IO_BASE + 0x04) /* 2 bytes */
-#define ACPI_PMA_CNT_BLK	(SB700_ACPI_IO_BASE + 0x0E) /* 1 byte */
-#define ACPI_PM_TMR_BLK		(SB700_ACPI_IO_BASE + 0x18) /* 4 bytes */
-#define ACPI_GPE0_BLK		(SB700_ACPI_IO_BASE + 0x10) /* 8 bytes */
+#define ACPI_PMA_CNT_BLK	(SB700_ACPI_IO_BASE + 0x16) /* 1 byte */
+#define ACPI_PM_TMR_BLK		(SB700_ACPI_IO_BASE + 0x20) /* 4 bytes */
+#define ACPI_GPE0_BLK		(SB700_ACPI_IO_BASE + 0x18) /* 8 bytes */
 #define ACPI_CPU_CONTROL	(SB700_ACPI_IO_BASE + 0x08) /* 6 bytes */
+#define ACPI_CPU_P_LVL2		(ACPI_CPU_CONTROL + 0x4)    /* 1 byte */
 
 extern void pm_iowrite(u8 reg, u8 value);
 extern u8 pm_ioread(u8 reg);
diff --git a/src/southbridge/amd/sb700/sm.c b/src/southbridge/amd/sb700/sm.c
index dc4b26fc14..04f4601883 100644
--- a/src/southbridge/amd/sb700/sm.c
+++ b/src/southbridge/amd/sb700/sm.c
@@ -123,7 +123,10 @@ static void sm_init(device_t dev)
 	pci_write_config8(dev, 0x41, byte);
 
 	byte = pm_ioread(0x61);
-	byte |= 1 << 1;		/* Set to enable NB/SB handshake during IOAPIC interrupt for AMD K8/K7 */
+	if (IS_ENABLED(CONFIG_CPU_AMD_MODEL_10XXX))
+		byte &= ~(1 << 1);	/* Clear for non-K8 CPUs */
+	else
+		byte |= 1 << 1;		/* Set to enable NB/SB handshake during IOAPIC interrupt for AMD K8/K7 */
 	pm_iowrite(0x61, byte);
 
 	/* disable SMI */
diff --git a/src/southbridge/amd/sb800/fadt.c b/src/southbridge/amd/sb800/fadt.c
index 763fd13e42..ce64036f6c 100644
--- a/src/southbridge/amd/sb800/fadt.c
+++ b/src/southbridge/amd/sb800/fadt.c
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2010 Advanced Micro Devices, Inc.
+ * Copyright (C) 2015 Raptor Engineering
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -22,6 +23,7 @@
 #include <arch/acpi.h>
 #include <arch/io.h>
 #include <device/device.h>
+#include <cpu/amd/powernow.h>
 #include "sb800.h"
 
 void acpi_create_fadt(acpi_fadt_t * fadt, acpi_facs_t * facs, void *dsdt)
@@ -152,5 +154,8 @@ void acpi_create_fadt(acpi_fadt_t * fadt, acpi_facs_t * facs, void *dsdt)
 	fadt->x_gpe1_blk.addrl = 0;
 	fadt->x_gpe1_blk.addrh = 0x0;
 
+	if (IS_ENABLED(CONFIG_CPU_AMD_MODEL_10XXX))
+		amd_powernow_update_fadt(fadt);
+
 	header->checksum = acpi_checksum((void *)fadt, sizeof(acpi_fadt_t));
 }
diff --git a/src/southbridge/amd/sb800/sb800.h b/src/southbridge/amd/sb800/sb800.h
index a30a447fa5..6a037cf4e8 100644
--- a/src/southbridge/amd/sb800/sb800.h
+++ b/src/southbridge/amd/sb800/sb800.h
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2010 Advanced Micro Devices, Inc.
+ * Copyright (C) 2015 Timothy Pearson <tpearson@raptorengineeringinc.com>, Raptor Engineering
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -31,10 +32,11 @@
 
 #define ACPI_PM_EVT_BLK		(SB800_ACPI_IO_BASE + 0x00) /* 4 bytes */
 #define ACPI_PM1_CNT_BLK	(SB800_ACPI_IO_BASE + 0x04) /* 2 bytes */
-#define ACPI_PMA_CNT_BLK	(SB800_ACPI_IO_BASE + 0x0F) /* 1 byte */
-#define ACPI_PM_TMR_BLK		(SB800_ACPI_IO_BASE + 0x18) /* 4 bytes */
-#define ACPI_GPE0_BLK		(SB800_ACPI_IO_BASE + 0x10) /* 8 bytes */
+#define ACPI_PMA_CNT_BLK	(SB800_ACPI_IO_BASE + 0x17) /* 1 byte */
+#define ACPI_PM_TMR_BLK		(SB800_ACPI_IO_BASE + 0x20) /* 4 bytes */
+#define ACPI_GPE0_BLK		(SB800_ACPI_IO_BASE + 0x18) /* 8 bytes */
 #define ACPI_CPU_CONTROL	(SB800_ACPI_IO_BASE + 0x08) /* 6 bytes */
+#define ACPI_CPU_P_LVL2		(ACPI_CPU_CONTROL + 0x4)    /* 1 byte */
 
 void pm_iowrite(u8 reg, u8 value);
 u8 pm_ioread(u8 reg);