Factor out a few commonly duplicated functions from northbridge.c.
[coreboot.git] / src / northbridge / amd / amdk8 / northbridge.c
index b94a674ba80e24e13d9b60e546605ea1b5948f6b..263777f75105243959238a4aca7ab27bb6e88c16 100644 (file)
@@ -1,3 +1,10 @@
+/* This should be done by Eric
+       2004.12 yhlu add dual core support
+       2005.01 yhlu add support move apic before pci_domain in MB Config.lb
+       2005.02 yhlu add e0 memory hole support
+       2005.11 yhlu add put sb ht chain on bus 0
+*/
+
 #include <console/console.h>
 #include <arch/io.h>
 #include <stdint.h>
 #include <string.h>
 #include <bitops.h>
 #include <cpu/cpu.h>
+
+#include <cpu/x86/lapic.h>
+
+#include <cpu/amd/multicore.h>
+#if CONFIG_LOGICAL_CPUS==1
+#include <pc80/mc146818rtc.h>
+#endif
+
 #include "chip.h"
 #include "root_complex/chip.h"
 #include "northbridge.h"
+
 #include "amdk8.h"
 
-#define FX_DEVS 8
-static device_t __f0_dev[FX_DEVS];
-static device_t __f1_dev[FX_DEVS];
+#include <cpu/amd/model_fxx_rev.h>
 
-#if 0
-static void debug_fx_devs(void)
-{
-       int i;
-       for (i = 0; i < FX_DEVS; i++) {
-               device_t dev;
-               dev = __f0_dev[i];
-               if (dev) {
-                       printk_debug("__f0_dev[%d]: %s bus: %p\n",
-                               i, dev_path(dev), dev->bus);
-               }
-               dev = __f1_dev[i];
-               if (dev) {
-                       printk_debug("__f1_dev[%d]: %s bus: %p\n",
-                               i, dev_path(dev), dev->bus);
-               }
-       }
-}
-#endif
+#include <cpu/amd/amdk8_sysconf.h>
+
+struct amdk8_sysconf_t sysconf;
+
+#define MAX_FX_DEVS 8
+static device_t __f0_dev[MAX_FX_DEVS];
+static device_t __f1_dev[MAX_FX_DEVS];
+static unsigned fx_devs=0;
 
 static void get_fx_devs(void)
 {
        int i;
-       if (__f1_dev[0]) {
-               return;
-       }
-       for (i = 0; i < FX_DEVS; i++) {
+       for(i = 0; i < MAX_FX_DEVS; i++) {
                __f0_dev[i] = dev_find_slot(0, PCI_DEVFN(0x18 + i, 0));
                __f1_dev[i] = dev_find_slot(0, PCI_DEVFN(0x18 + i, 1));
+               if (__f0_dev[i] != NULL && __f1_dev[i] != NULL)
+                       fx_devs = i+1;
        }
-       if (!__f1_dev[0]) {
-               die("Cannot find 0:0x18.1\n");
+       if (__f1_dev[0] == NULL || __f0_dev[0] == NULL || fx_devs == 0) {
+               die("Cannot find 0:0x18.[0|1]\n");
        }
 }
 
-static uint32_t f1_read_config32(unsigned reg)
+static u32 f1_read_config32(unsigned reg)
 {
-       get_fx_devs();
+       if (fx_devs == 0)
+               get_fx_devs();
        return pci_read_config32(__f1_dev[0], reg);
 }
 
-static void f1_write_config32(unsigned reg, uint32_t value)
+static void f1_write_config32(unsigned reg, u32 value)
 {
        int i;
-       get_fx_devs();
-       for (i = 0; i < FX_DEVS; i++) {
+       if (fx_devs == 0)
+               get_fx_devs();
+       for(i = 0; i < fx_devs; i++) {
                device_t dev;
                dev = __f1_dev[i];
                if (dev && dev->enabled) {
@@ -72,50 +76,51 @@ static void f1_write_config32(unsigned reg, uint32_t value)
        }
 }
 
-static unsigned int amdk8_nodeid(device_t dev)
+static u32 amdk8_nodeid(device_t dev)
 {
-       return (dev->path.u.pci.devfn >> 3) - 0x18;
+       return (dev->path.pci.devfn >> 3) - 0x18;
 }
 
-static unsigned int amdk8_scan_chains(device_t dev, unsigned int max)
+static u32 amdk8_scan_chain(device_t dev, u32 nodeid, struct bus *link, u32 link_num, u32 sblink,
+                               u32 max, u32 offset_unitid)
 {
-       unsigned nodeid;
-       unsigned link;
-       nodeid = amdk8_nodeid(dev);
-#if 0
-       printk_debug("%s amdk8_scan_chains max: %d starting...\n", 
-                    dev_path(dev), max);
-#endif
-       for (link = 0; link < dev->links; link++) {
-               uint32_t link_type;
-               uint32_t busses, config_busses;
-               unsigned free_reg, config_reg;
-               dev->link[link].cap = 0x80 + (link *0x20);
+
+               u32 link_type;
+               int i;
+               u32 busses, config_busses;
+               u32 free_reg, config_reg;
+               u32 ht_unitid_base[4]; // here assume only 4 HT device on chain
+               u32 max_bus;
+               u32 min_bus;
+               u32 max_devfn;
+
+               link->cap = 0x80 + (link_num *0x20);
                do {
-                       link_type = pci_read_config32(dev, dev->link[link].cap + 0x18);
+                       link_type = pci_read_config32(dev, link->cap + 0x18);
                } while(link_type & ConnectionPending);
                if (!(link_type & LinkConnected)) {
-                       continue;
+                       return max;
                }
                do {
-                       link_type = pci_read_config32(dev, dev->link[link].cap + 0x18);
+                       link_type = pci_read_config32(dev, link->cap + 0x18);
                } while(!(link_type & InitComplete));
                if (!(link_type & NonCoherent)) {
-                       continue;
+                       return max;
                }
                /* See if there is an available configuration space mapping
-                * register in function 1. */
+                * register in function 1.
+                */
                free_reg = 0;
-               for (config_reg = 0xe0; config_reg <= 0xec; config_reg += 4) {
-                       uint32_t config;
+               for(config_reg = 0xe0; config_reg <= 0xec; config_reg += 4) {
+                       u32 config;
                        config = f1_read_config32(config_reg);
                        if (!free_reg && ((config & 3) == 0)) {
                                free_reg = config_reg;
                                continue;
                        }
-                       if (((config & 3) == 3) && 
-                           (((config >> 4) & 7) == nodeid) &&
-                           (((config >> 8) & 3) == link)) {
+                       if (((config & 3) == 3) &&
+                               (((config >> 4) & 7) == nodeid) &&
+                               (((config >> 8) & 3) == link_num)) {
                                break;
                        }
                }
@@ -123,22 +128,45 @@ static unsigned int amdk8_scan_chains(device_t dev, unsigned int max)
                        config_reg = free_reg;
                }
                /* If we can't find an available configuration space mapping
-                * register skip this bus */
+                * register skip this bus
+                */
                if (config_reg > 0xec) {
-                       continue;
+                       return max;
                }
 
                /* Set up the primary, secondary and subordinate bus numbers.
                 * We have no idea how many busses are behind this bridge yet,
                 * so we set the subordinate bus number to 0xff for the moment.
                 */
-               dev->link[link].secondary = ++max;
-               dev->link[link].subordinate = 0xff;
+#if CONFIG_SB_HT_CHAIN_ON_BUS0 > 0
+               // first chain will on bus 0
+               if((nodeid == 0) && (sblink==link_num)) { // actually max is 0 here
+                       min_bus = max;
+               }
+       #if CONFIG_SB_HT_CHAIN_ON_BUS0 > 1
+               // second chain will be on 0x40, third 0x80, forth 0xc0
+               else {
+                       min_bus = ((max>>6) + 1) * 0x40;
+               }
+               max = min_bus;
+       #else
+               //other ...
+               else {
+                       min_bus = ++max;
+               }
+       #endif
+#else
+               min_bus = ++max;
+#endif
+               max_bus = 0xff;
+
+               link->secondary = min_bus;
+               link->subordinate = max_bus;
 
                /* Read the existing primary/secondary/subordinate bus
                 * number configuration.
                 */
-               busses = pci_read_config32(dev, dev->link[link].cap + 0x14);
+               busses = pci_read_config32(dev, link->cap + 0x14);
                config_busses = f1_read_config32(config_reg);
 
                /* Configure the bus numbers for this bridge: the configuration
@@ -147,92 +175,135 @@ static unsigned int amdk8_scan_chains(device_t dev, unsigned int max)
                 */
                busses &= 0xff000000;
                busses |= (((unsigned int)(dev->bus->secondary) << 0) |
-                          ((unsigned int)(dev->link[link].secondary) << 8) |
-                          ((unsigned int)(dev->link[link].subordinate) << 16));
-               pci_write_config32(dev, dev->link[link].cap + 0x14, busses);
+                       ((unsigned int)(link->secondary) << 8) |
+                       ((unsigned int)(link->subordinate) << 16));
+               pci_write_config32(dev, link->cap + 0x14, busses);
 
                config_busses &= 0x000fc88;
-               config_busses |= 
+               config_busses |=
                        (3 << 0) |  /* rw enable, no device compare */
-                       (( nodeid & 7) << 4) | 
-                       (( link & 3 ) << 8) |  
-                       ((dev->link[link].secondary) << 16) |
-                       ((dev->link[link].subordinate) << 24);
+                       (( nodeid & 7) << 4) |
+                       (( link_num & 3 ) << 8) |
+                       ((link->secondary) << 16) |
+                       ((link->subordinate) << 24);
                f1_write_config32(config_reg, config_busses);
 
-#if 0
-               printk_debug("%s Hyper transport scan link: %d max: %d\n", 
-                       dev_path(dev), link, max);
-#endif
                /* Now we can scan all of the subordinate busses i.e. the
-                * chain on the hypertranport link */
-               max = hypertransport_scan_chain(&dev->link[link], max);
+                * chain on the hypertranport link
+                */
+               for(i=0;i<4;i++) {
+                       ht_unitid_base[i] = 0x20;
+               }
 
-#if 0
-               printk_debug("%s Hyper transport scan link: %d new max: %d\n",
-                       dev_path(dev), link, max);
-#endif
+               if (min_bus == 0)
+                       max_devfn = (0x17<<3) | 7;
+               else
+                       max_devfn = (0x1f<<3) | 7;
+
+               max = hypertransport_scan_chain(link, 0, max_devfn, max, ht_unitid_base, offset_unitid);
 
                /* We know the number of busses behind this bridge.  Set the
                 * subordinate bus number to it's real value
                 */
-               dev->link[link].subordinate = max;
+               link->subordinate = max;
                busses = (busses & 0xff00ffff) |
-                       ((unsigned int) (dev->link[link].subordinate) << 16);
-               pci_write_config32(dev, dev->link[link].cap + 0x14, busses);
+                       ((unsigned int) (link->subordinate) << 16);
+               pci_write_config32(dev, link->cap + 0x14, busses);
 
                config_busses = (config_busses & 0x00ffffff) |
-                       (dev->link[link].subordinate << 24);
+                       (link->subordinate << 24);
                f1_write_config32(config_reg, config_busses);
-#if 0
-               printk_debug("%s Hypertransport scan link: %d done\n",
-                       dev_path(dev), link);
+
+               {
+                       // use config_reg and ht_unitid_base to update hcdn_reg
+                       int index;
+                       u32 temp = 0;
+                       index = (config_reg-0xe0) >> 2;
+                       for(i=0;i<4;i++) {
+                               temp |= (ht_unitid_base[i] & 0xff) << (i*8);
+                       }
+
+                       sysconf.hcdn_reg[index] = temp;
+
+               }
+       return max;
+}
+
+static unsigned amdk8_scan_chains(device_t dev, unsigned max)
+{
+       unsigned nodeid;
+       struct bus *link;
+       unsigned sblink = 0;
+       unsigned offset_unitid = 0;
+
+       nodeid = amdk8_nodeid(dev);
+
+       if(nodeid==0) {
+               sblink = (pci_read_config32(dev, 0x64)>>8) & 3;
+#if CONFIG_SB_HT_CHAIN_ON_BUS0 > 0
+       #if ((CONFIG_HT_CHAIN_UNITID_BASE != 1) || (CONFIG_HT_CHAIN_END_UNITID_BASE != 0x20))
+               offset_unitid = 1;
+       #endif
+               for (link = dev->link_list; link; link = link->next)
+                       if (link->link_num == sblink)
+                               max = amdk8_scan_chain(dev, nodeid, link, sblink, sblink, max, offset_unitid ); // do sb ht chain at first, in case s2885 put sb chain (8131/8111) on link2, but put 8151 on link0
 #endif
        }
-#if 0
-       printk_debug("%s amdk8_scan_chains max: %d done\n", 
-               dev_path(dev), max);
+
+       for (link = dev->link_list; link; link = link->next) {
+#if CONFIG_SB_HT_CHAIN_ON_BUS0 > 0
+               if( (nodeid == 0) && (sblink == link->link_num) ) continue; //already done
 #endif
+               offset_unitid = 0;
+               #if ((CONFIG_HT_CHAIN_UNITID_BASE != 1) || (CONFIG_HT_CHAIN_END_UNITID_BASE != 0x20))
+                       #if CONFIG_SB_HT_CHAIN_UNITID_OFFSET_ONLY == 1
+                       if((nodeid == 0) && (sblink == link->link_num))
+                       #endif
+                               offset_unitid = 1;
+               #endif
+
+               max = amdk8_scan_chain(dev, nodeid, link, link->link_num, sblink, max, offset_unitid);
+       }
        return max;
 }
 
+
 static int reg_useable(unsigned reg, device_t goal_dev, unsigned goal_nodeid,
-                      unsigned goal_link)
+                       unsigned goal_link)
 {
        struct resource *res;
-       unsigned nodeid, link;
+       unsigned nodeid, link = 0;
        int result;
        res = 0;
-       for (nodeid = 0; !res && (nodeid < 8); nodeid++) {
+       for(nodeid = 0; !res && (nodeid < fx_devs); nodeid++) {
                device_t dev;
                dev = __f0_dev[nodeid];
-               for (link = 0; !res && (link < 3); link++) {
-                       res = probe_resource(dev, 0x100 + (reg | link));
+               if (!dev)
+                       continue;
+               for(link = 0; !res && (link < 3); link++) {
+                       res = probe_resource(dev, IOINDEX(0x100 + reg, link));
                }
        }
        result = 2;
        if (res) {
                result = 0;
-               if ((goal_link == (link - 1)) && 
-                   (goal_nodeid == (nodeid - 1)) &&
-                   (res->flags <= 1)) {
+               if (    (goal_link == (link - 1)) &&
+                       (goal_nodeid == (nodeid - 1)) &&
+                       (res->flags <= 1)) {
                        result = 1;
                }
        }
-#if 0
-       printk_debug("reg: %02x result: %d gnodeid: %u glink: %u nodeid: %u link: %u\n",
-                    reg, result, goal_nodeid, goal_link, nodeid, link);
-#endif
        return result;
 }
 
-static struct resource *amdk8_find_iopair(device_t dev, unsigned nodeid, unsigned link)
+static unsigned amdk8_find_reg(device_t dev, unsigned nodeid, unsigned link,
+                              unsigned min, unsigned max)
 {
-       struct resource *resource;
+       unsigned resource;
        unsigned free_reg, reg;
        resource = 0;
        free_reg = 0;
-       for (reg = 0xc0; reg <= 0xd8; reg += 0x8) {
+       for(reg = min; reg <= max; reg += 0x8) {
                int result;
                result = reg_useable(reg, dev, nodeid, link);
                if (result == 1) {
@@ -244,115 +315,100 @@ static struct resource *amdk8_find_iopair(device_t dev, unsigned nodeid, unsigne
                        free_reg = reg;
                }
        }
-       if (reg > 0xd8) {
+       if (reg > max) {
                reg = free_reg;
        }
        if (reg > 0) {
-               resource = new_resource(dev, 0x100 + (reg | link));
+               resource = IOINDEX(0x100 + reg, link);
        }
        return resource;
 }
 
-static struct resource *amdk8_find_mempair(device_t dev, unsigned nodeid, unsigned link)
+static unsigned amdk8_find_iopair(device_t dev, unsigned nodeid, unsigned link)
 {
-       struct resource *resource;
-       unsigned free_reg, reg;
-       resource = 0;
-       free_reg = 0;
-       for (reg = 0x80; reg <= 0xb8; reg += 0x8) {
-               int result;
-               result = reg_useable(reg, dev, nodeid, link);
-               if (result == 1) {
-                       /* I have been allocated this one */
-                       break;
-               }
-               else if (result > 1) {
-                       /* I have a free register pair */
-                       free_reg = reg;
-               }
-       }
-       if (reg > 0xb8) {
-               reg = free_reg;
-       }
-       if (reg > 0) {
-               resource = new_resource(dev, 0x100 + (reg | link));
-       }
-       return resource;
+       return amdk8_find_reg(dev, nodeid, link, 0xc0, 0xd8);
+}
+
+static unsigned amdk8_find_mempair(device_t dev, unsigned nodeid, unsigned link)
+{
+       return amdk8_find_reg(dev, nodeid, link, 0x80, 0xb8);
 }
 
 static void amdk8_link_read_bases(device_t dev, unsigned nodeid, unsigned link)
 {
        struct resource *resource;
-       
+
        /* Initialize the io space constraints on the current bus */
-       resource =  amdk8_find_iopair(dev, nodeid, link);
+       resource = new_resource(dev, IOINDEX(0, link));
        if (resource) {
                resource->base  = 0;
                resource->size  = 0;
                resource->align = log2(HT_IO_HOST_ALIGN);
                resource->gran  = log2(HT_IO_HOST_ALIGN);
                resource->limit = 0xffffUL;
-               resource->flags = IORESOURCE_IO;
-               compute_allocate_resource(&dev->link[link], resource, 
-                       IORESOURCE_IO, IORESOURCE_IO);
+               resource->flags = IORESOURCE_IO | IORESOURCE_BRIDGE;
        }
 
        /* Initialize the prefetchable memory constraints on the current bus */
-       resource = amdk8_find_mempair(dev, nodeid, link);
+       resource = new_resource(dev, IOINDEX(2, link));
        if (resource) {
-               resource->base  = 0;
-               resource->size  = 0;
+               resource->base = 0;
+               resource->size = 0;
                resource->align = log2(HT_MEM_HOST_ALIGN);
-               resource->gran  = log2(HT_MEM_HOST_ALIGN);
+               resource->gran = log2(HT_MEM_HOST_ALIGN);
                resource->limit = 0xffffffffffULL;
                resource->flags = IORESOURCE_MEM | IORESOURCE_PREFETCH;
-               compute_allocate_resource(&dev->link[link], resource, 
-                       IORESOURCE_MEM | IORESOURCE_PREFETCH, 
-                       IORESOURCE_MEM | IORESOURCE_PREFETCH);
+               resource->flags |= IORESOURCE_BRIDGE;
        }
 
        /* Initialize the memory constraints on the current bus */
-       resource = amdk8_find_mempair(dev, nodeid, link);
+       resource = new_resource(dev, IOINDEX(1, link));
        if (resource) {
-               resource->base  = 0;
-               resource->size  = 0;
+               resource->base = 0;
+               resource->size = 0;
                resource->align = log2(HT_MEM_HOST_ALIGN);
-               resource->gran  = log2(HT_MEM_HOST_ALIGN);
-               resource->limit = 0xffffffffffULL;
-               resource->flags = IORESOURCE_MEM;
-               compute_allocate_resource(&dev->link[link], resource, 
-                       IORESOURCE_MEM | IORESOURCE_PREFETCH, 
-                       IORESOURCE_MEM);
+               resource->gran = log2(HT_MEM_HOST_ALIGN);
+               resource->limit = 0xffffffffULL;
+               resource->flags = IORESOURCE_MEM | IORESOURCE_BRIDGE;
        }
 }
 
+static void amdk8_create_vga_resource(device_t dev, unsigned nodeid);
+
 static void amdk8_read_resources(device_t dev)
 {
-       unsigned nodeid, link;
+       unsigned nodeid;
+       struct bus *link;
        nodeid = amdk8_nodeid(dev);
-       for (link = 0; link < dev->links; link++) {
-               if (dev->link[link].children) {
-                       amdk8_link_read_bases(dev, nodeid, link);
+       for(link = dev->link_list; link; link = link->next) {
+               if (link->children) {
+                       amdk8_link_read_bases(dev, nodeid, link->link_num);
                }
        }
+       amdk8_create_vga_resource(dev, nodeid);
 }
 
 static void amdk8_set_resource(device_t dev, struct resource *resource, unsigned nodeid)
 {
+       struct bus *link;
        resource_t rbase, rend;
-       unsigned reg, link;
+       unsigned reg, link_num;
        char buf[50];
 
        /* Make certain the resource has actually been set */
        if (!(resource->flags & IORESOURCE_ASSIGNED)) {
+               printk(BIOS_ERR, "%s: can't set unassigned resource @%lx %lx\n",
+                          __func__, resource->index, resource->flags);
                return;
        }
 
        /* If I have already stored this resource don't worry about it */
        if (resource->flags & IORESOURCE_STORED) {
+               printk(BIOS_ERR, "%s: can't set stored resource @%lx %lx\n", __func__,
+                          resource->index, resource->flags);
                return;
        }
-       
+
        /* Only handle PCI memory and IO resources */
        if (!(resource->flags & (IORESOURCE_MEM | IORESOURCE_IO)))
                return;
@@ -361,20 +417,32 @@ static void amdk8_set_resource(device_t dev, struct resource *resource, unsigned
        if (resource->index < 0x100) {
                return;
        }
+
+       if (resource->size == 0)
+               return;
+
        /* Get the base address */
        rbase = resource->base;
-       
+
        /* Get the limit (rounded up) */
        rend  = resource_end(resource);
 
        /* Get the register and link */
        reg  = resource->index & 0xfc;
-       link = resource->index & 3;
+       link_num = IOINDEX_LINK(resource->index);
+
+       for (link = dev->link_list; link; link = link->next)
+               if (link->link_num == link_num)
+                       break;
+
+       if (link == NULL) {
+               printk(BIOS_ERR, "%s: can't find link %x for %lx\n", __func__,
+                          link_num, resource->index);
+               return;
+       }
 
        if (resource->flags & IORESOURCE_IO) {
-               uint32_t base, limit;
-               compute_allocate_resource(&dev->link[link], resource,
-                       IORESOURCE_IO, IORESOURCE_IO);
+               u32 base, limit;
                base  = f1_read_config32(reg);
                limit = f1_read_config32(reg + 0x4);
                base  &= 0xfe000fcc;
@@ -382,24 +450,23 @@ static void amdk8_set_resource(device_t dev, struct resource *resource, unsigned
                base  |= 3;
                limit &= 0xfe000fc8;
                limit |= rend & 0x01fff000;
-               limit |= (link & 3) << 4;
+               limit |= (link_num & 3) << 4;
                limit |= (nodeid & 7);
 
-               if (dev->link[link].bridge_ctrl & PCI_BRIDGE_CTL_VGA) {
+               if (link->bridge_ctrl & PCI_BRIDGE_CTL_VGA) {
+                       printk(BIOS_SPEW, "%s, enabling legacy VGA IO forwarding for %s link 0x%x\n",
+                                   __func__, dev_path(dev), link_num);
                        base |= PCI_IO_BASE_VGA_EN;
                }
-               if (dev->link[link].bridge_ctrl & PCI_BRIDGE_CTL_NO_ISA) {
+               if (link->bridge_ctrl & PCI_BRIDGE_CTL_NO_ISA) {
                        base |= PCI_IO_BASE_NO_ISA;
                }
-               
+
                f1_write_config32(reg + 0x4, limit);
                f1_write_config32(reg, base);
        }
        else if (resource->flags & IORESOURCE_MEM) {
-               uint32_t base, limit;
-               compute_allocate_resource(&dev->link[link], resource,
-                       IORESOURCE_MEM | IORESOURCE_PREFETCH,
-                       resource->flags & (IORESOURCE_MEM | IORESOURCE_PREFETCH));
+               u32 base, limit;
                base  = f1_read_config32(reg);
                limit = f1_read_config32(reg + 0x4);
                base  &= 0x000000f0;
@@ -407,220 +474,196 @@ static void amdk8_set_resource(device_t dev, struct resource *resource, unsigned
                base  |= 3;
                limit &= 0x00000048;
                limit |= (rend >> 8) & 0xffffff00;
-               limit |= (link & 3) << 4;
+               limit |= (link_num & 3) << 4;
                limit |= (nodeid & 7);
                f1_write_config32(reg + 0x4, limit);
                f1_write_config32(reg, base);
        }
        resource->flags |= IORESOURCE_STORED;
-       sprintf(buf, " <node %d link %d>",
-               nodeid, link);
+       sprintf(buf, " <node %x link %x>",
+               nodeid, link_num);
        report_resource_stored(dev, resource, buf);
 }
 
-/**
- *
- * I tried to reuse the resource allocation code in amdk8_set_resource()
- * but it is too diffcult to deal with the resource allocation magic.
- */
+#if CONFIG_CONSOLE_VGA_MULTI == 1
+extern device_t vga_pri;       // the primary vga device, defined in device.c
+#endif
+
 static void amdk8_create_vga_resource(device_t dev, unsigned nodeid)
 {
        struct resource *resource;
-       unsigned link;
-       uint32_t base, limit;
-       unsigned reg;
+       struct bus *link;
 
        /* find out which link the VGA card is connected,
         * we only deal with the 'first' vga card */
-       for (link = 0; link < dev->links; link++) {
-               if (dev->link[link].bridge_ctrl & PCI_BRIDGE_CTL_VGA) {
+       for (link = dev->link_list; link; link = link->next) {
+               if (link->bridge_ctrl & PCI_BRIDGE_CTL_VGA) {
+#if CONFIG_CONSOLE_VGA_MULTI == 1
+                       printk(BIOS_DEBUG, "VGA: vga_pri bus num = %d link bus range [%d,%d]\n", vga_pri->bus->secondary,
+                               link->secondary,link->subordinate);
+                       /* We need to make sure the vga_pri is under the link */
+                       if((vga_pri->bus->secondary >= link->secondary ) &&
+                               (vga_pri->bus->secondary <= link->subordinate )
+                       )
+#endif
                        break;
                }
        }
 
        /* no VGA card installed */
-       if (link == dev->links)
+       if (link == NULL)
                return;
 
-       /* allocate a temp resrouce for legacy VGA buffer */
-       resource = amdk8_find_mempair(dev, nodeid, link);
+       printk(BIOS_DEBUG, "VGA: %s (aka node %d) link %d has VGA device\n", dev_path(dev), nodeid, link->link_num);
+
+       /* allocate a temp resource for the legacy VGA buffer */
+       resource = new_resource(dev, IOINDEX(4, link->link_num));
+       if(!resource){
+               printk(BIOS_DEBUG, "VGA: %s out of resources.\n", dev_path(dev));
+               return;
+       }
        resource->base = 0xa0000;
        resource->size = 0x20000;
-
-       /* write the resource to the hardware */
-       reg  = resource->index & 0xfc;
-       base  = f1_read_config32(reg);
-       limit = f1_read_config32(reg + 0x4);
-       base  &= 0x000000f0;
-       base  |= (resource->base >> 8) & 0xffffff00;
-       base  |= 3;
-       limit &= 0x00000048;
-       limit |= ((resource->base + resource->size) >> 8) & 0xffffff00;
-       limit |= (resource->index & 3) << 4;
-       limit |= (nodeid & 7);
-       f1_write_config32(reg + 0x4, limit);
-       f1_write_config32(reg, base);
-
-       /* release the temp resource */
-       resource->flags = 0;
+       resource->limit = 0xffffffff;
+       resource->flags = IORESOURCE_FIXED | IORESOURCE_MEM |
+                         IORESOURCE_ASSIGNED;
 }
 
 static void amdk8_set_resources(device_t dev)
 {
-       unsigned nodeid, link;
-       int i;
+       unsigned nodeid;
+       struct bus *bus;
+       struct resource *res;
 
        /* Find the nodeid */
        nodeid = amdk8_nodeid(dev);
 
-       amdk8_create_vga_resource(dev, nodeid);
-       
        /* Set each resource we have found */
-       for (i = 0; i < dev->resources; i++) {
-               amdk8_set_resource(dev, &dev->resource[i], nodeid);
+       for(res = dev->resource_list; res; res = res->next) {
+               struct resource *old = NULL;
+               unsigned index;
+
+               if (res->size == 0) /* No need to allocate registers. */
+                       continue;
+
+               if (res->flags & IORESOURCE_IO)
+                       index = amdk8_find_iopair(dev, nodeid,
+                                                 IOINDEX_LINK(res->index));
+               else
+                       index = amdk8_find_mempair(dev, nodeid,
+                                                  IOINDEX_LINK(res->index));
+
+               old = probe_resource(dev, index);
+               if (old) {
+                       res->index = old->index;
+                       old->index = 0;
+                       old->flags = 0;
+               }
+               else
+                       res->index = index;
+
+               amdk8_set_resource(dev, res, nodeid);
        }
 
-       for (link = 0; link < dev->links; link++) {
-               struct bus *bus;
-               bus = &dev->link[link];
+       compact_resources(dev);
+
+       for(bus = dev->link_list; bus; bus = bus->next) {
                if (bus->children) {
                        assign_resources(bus);
                }
        }
 }
 
-static void amdk8_enable_resources(device_t dev)
-{
-       pci_dev_enable_resources(dev);
-       enable_childrens_resources(dev);
-}
-
 static void mcf0_control_init(struct device *dev)
 {
-       uint32_t cmd;
-
-#if 0  
-       printk_debug("NB: Function 0 Misc Control.. ");
-#endif
-#if 1
-       /* improve latency and bandwith on HT */
-       cmd = pci_read_config32(dev, 0x68);
-       cmd &= 0xffff80ff;
-       cmd |= 0x00004800;
-       pci_write_config32(dev, 0x68, cmd );
+#if 0
+       printk(BIOS_DEBUG, "NB: Function 0 Misc Control.. ");
 #endif
-
-#if 0  
-       /* over drive the ht port to 1000 Mhz */
-       cmd = pci_read_config32(dev, 0xa8);
-       cmd &= 0xfffff0ff;
-       cmd |= 0x00000600;
-       pci_write_config32(dev, 0xdc, cmd );
-#endif 
 #if 0
-       printk_debug("done.\n");
+       printk(BIOS_DEBUG, "done.\n");
 #endif
 }
 
 static struct device_operations northbridge_operations = {
-       .read_resources   = amdk8_read_resources,
-       .set_resources    = amdk8_set_resources,
-       .enable_resources = amdk8_enable_resources,
-       .init             = mcf0_control_init,
-       .scan_bus         = amdk8_scan_chains,
-       .enable           = 0,
-       .ops_pci          = 0,
+       .read_resources   = amdk8_read_resources,
+       .set_resources    = amdk8_set_resources,
+       .enable_resources = pci_dev_enable_resources,
+       .init             = mcf0_control_init,
+       .scan_bus         = amdk8_scan_chains,
+       .enable           = 0,
+       .ops_pci          = 0,
 };
 
 
-static struct pci_driver mcf0_driver __pci_driver = {
-       .ops    = &northbridge_operations,
+static const struct pci_driver mcf0_driver __pci_driver = {
+       .ops    = &northbridge_operations,
        .vendor = PCI_VENDOR_ID_AMD,
        .device = 0x1100,
 };
 
-#if CONFIG_CHIP_NAME == 1
-
 struct chip_operations northbridge_amd_amdk8_ops = {
        CHIP_NAME("AMD K8 Northbridge")
        .enable_dev = 0,
 };
 
-#endif
-
-static void pci_domain_read_resources(device_t dev)
+static void amdk8_domain_read_resources(device_t dev)
 {
-       struct resource *resource;
        unsigned reg;
 
        /* Find the already assigned resource pairs */
        get_fx_devs();
-       for (reg = 0x80; reg <= 0xd8; reg+= 0x08) {
-               uint32_t base, limit;
+       for(reg = 0x80; reg <= 0xd8; reg+= 0x08) {
+               u32 base, limit;
                base  = f1_read_config32(reg);
                limit = f1_read_config32(reg + 0x04);
                /* Is this register allocated? */
                if ((base & 3) != 0) {
-                       unsigned nodeid, link;
-                       device_t dev;
+                       unsigned nodeid, reg_link;
+                       device_t reg_dev;
                        nodeid = limit & 7;
-                       link   = (limit >> 4) & 3;
-                       dev = __f0_dev[nodeid];
-                       if (dev) {
+                       reg_link = (limit >> 4) & 3;
+                       reg_dev = __f0_dev[nodeid];
+                       if (reg_dev) {
                                /* Reserve the resource  */
-                               struct resource *resource;
-                               resource = new_resource(dev, 0x100 + (reg | link));
-                               if (resource) {
-                                       resource->flags = 1;
+                               struct resource *res;
+                               res = new_resource(reg_dev, IOINDEX(0x100 + reg, reg_link));
+                               if (res) {
+                                       res->base = base;
+                                       res->limit = limit;
+                                       res->flags = 1;
                                }
                        }
                }
        }
 
-       /* Initialize the system wide io space constraints */
-       resource = new_resource(dev, IOINDEX_SUBTRACTIVE(0, 0));
-       resource->base  = 0x400;
-       resource->limit = 0xffffUL;
-       resource->flags = IORESOURCE_IO | IORESOURCE_SUBTRACTIVE | IORESOURCE_ASSIGNED;
+       pci_domain_read_resources(dev);
 
-       /* Initialize the system wide memory resources constraints */
-       resource = new_resource(dev, IOINDEX_SUBTRACTIVE(1, 0));
+#if CONFIG_PCI_64BIT_PREF_MEM == 1
+       /* Initialize the system wide prefetchable memory resources constraints */
+       resource = new_resource(dev, 2);
        resource->limit = 0xfcffffffffULL;
-       resource->flags = IORESOURCE_MEM | IORESOURCE_SUBTRACTIVE | IORESOURCE_ASSIGNED;
-}
-
-static void ram_resource(device_t dev, unsigned long index,
-                        unsigned long basek, unsigned long sizek)
-{
-       struct resource *resource;
-
-       if (!sizek) {
-               return;
-       }
-       resource = new_resource(dev, index);
-       resource->base  = ((resource_t)basek) << 10;
-       resource->size  = ((resource_t)sizek) << 10;
-       resource->flags =  IORESOURCE_MEM | IORESOURCE_CACHEABLE | \
-               IORESOURCE_FIXED | IORESOURCE_STORED | IORESOURCE_ASSIGNED;
+       resource->flags = IORESOURCE_MEM | IORESOURCE_PREFETCH;
+#endif
 }
 
-static void tolm_test(void *gp, struct device *dev, struct resource *new)
+static void my_tolm_test(void *gp, struct device *dev, struct resource *new)
 {
        struct resource **best_p = gp;
        struct resource *best;
        best = *best_p;
-       if (!best || (best->base > new->base)) {
+       /* Skip VGA. */
+       if (!best || (best->base > new->base && new->base > 0xa0000)) {
                best = new;
        }
        *best_p = best;
 }
 
-static uint32_t find_pci_tolm(struct bus *bus)
+static u32 my_find_pci_tolm(struct bus *bus)
 {
        struct resource *min;
-       uint32_t tolm;
+       u32 tolm;
        min = 0;
-       search_bus_resources(bus, IORESOURCE_MEM, IORESOURCE_MEM, tolm_test, &min);
+       search_bus_resources(bus, IORESOURCE_MEM, IORESOURCE_MEM, my_tolm_test, &min);
        tolm = 0xffffffffUL;
        if (min && tolm > min->base) {
                tolm = min->base;
@@ -628,29 +671,304 @@ static uint32_t find_pci_tolm(struct bus *bus)
        return tolm;
 }
 
-static void pci_domain_set_resources(device_t dev)
+#if CONFIG_HW_MEM_HOLE_SIZEK != 0
+
+struct hw_mem_hole_info {
+       unsigned hole_startk;
+       int node_id;
+};
+
+static struct hw_mem_hole_info get_hw_mem_hole_info(void)
 {
+               struct hw_mem_hole_info mem_hole;
+               int i;
+
+               mem_hole.hole_startk = CONFIG_HW_MEM_HOLE_SIZEK;
+               mem_hole.node_id = -1;
+
+               for (i = 0; i < fx_devs; i++) {
+                       u32 base;
+                       u32 hole;
+                       base  = f1_read_config32(0x40 + (i << 3));
+                       if ((base & ((1<<1)|(1<<0))) != ((1<<1)|(1<<0))) {
+                               continue;
+                       }
+
+                       hole = pci_read_config32(__f1_dev[i], 0xf0);
+                       if(hole & 1) { // we find the hole
+                               mem_hole.hole_startk = (hole & (0xff<<24)) >> 10;
+                               mem_hole.node_id = i; // record the node No with hole
+                               break; // only one hole
+                       }
+               }
+
+               //We need to double check if there is speical set on base reg and limit reg are not continous instead of hole, it will find out it's hole_startk
+               if(mem_hole.node_id==-1) {
+                       u32 limitk_pri = 0;
+                       for(i=0; i<8; i++) {
+                               u32 base, limit;
+                               unsigned base_k, limit_k;
+                               base  = f1_read_config32(0x40 + (i << 3));
+                               if ((base & ((1<<1)|(1<<0))) != ((1<<1)|(1<<0))) {
+                                       continue;
+                               }
+
+                               base_k = (base & 0xffff0000) >> 2;
+                               if(limitk_pri != base_k) { // we find the hole
+                                       mem_hole.hole_startk = limitk_pri;
+                                       mem_hole.node_id = i;
+                                       break; //only one hole
+                               }
+
+                               limit = f1_read_config32(0x44 + (i << 3));
+                               limit_k = ((limit + 0x00010000) & 0xffff0000) >> 2;
+                               limitk_pri = limit_k;
+                       }
+               }
+               return mem_hole;
+}
+
+static void disable_hoist_memory(unsigned long hole_startk, int node_id)
+{
+       int i;
+       device_t dev;
+       u32 base, limit;
+       u32 hoist;
+       u32 hole_sizek;
+
+
+       //1. find which node has hole
+       //2. change limit in that node.
+       //3. change base and limit in later node
+       //4. clear that node f0
+
+       //if there is not mem hole enabled, we need to change it's base instead
+
+       hole_sizek = (4*1024*1024) - hole_startk;
+
+       for(i=7;i>node_id;i--) {
+
+               base  = f1_read_config32(0x40 + (i << 3));
+               if ((base & ((1<<1)|(1<<0))) != ((1<<1)|(1<<0))) {
+                       continue;
+               }
+               limit = f1_read_config32(0x44 + (i << 3));
+               f1_write_config32(0x44 + (i << 3),limit - (hole_sizek << 2));
+               f1_write_config32(0x40 + (i << 3),base - (hole_sizek << 2));
+       }
+       limit = f1_read_config32(0x44 + (node_id << 3));
+       f1_write_config32(0x44 + (node_id << 3),limit - (hole_sizek << 2));
+       dev = __f1_dev[node_id];
+       if (dev == NULL) {
+               printk(BIOS_ERR, "%s: node %x is NULL!\n", __func__, node_id);
+               return;
+       }
+       hoist = pci_read_config32(dev, 0xf0);
+       if(hoist & 1) {
+               pci_write_config32(dev, 0xf0, 0);
+       } else {
+               base = pci_read_config32(dev, 0x40 + (node_id << 3));
+               f1_write_config32(0x40 + (node_id << 3),base - (hole_sizek << 2));
+       }
+}
+
+static u32 hoist_memory(unsigned long hole_startk, int node_id)
+{
+       int i;
+       u32 carry_over;
+       device_t dev;
+       u32 base, limit;
+       u32 basek;
+       u32 hoist;
+
+       carry_over = (4*1024*1024) - hole_startk;
+
+       for(i=7;i>node_id;i--) {
+
+               base  = f1_read_config32(0x40 + (i << 3));
+               if ((base & ((1<<1)|(1<<0))) != ((1<<1)|(1<<0))) {
+                       continue;
+               }
+               limit = f1_read_config32(0x44 + (i << 3));
+               f1_write_config32(0x44 + (i << 3),limit + (carry_over << 2));
+               f1_write_config32(0x40 + (i << 3),base + (carry_over << 2));
+       }
+       limit = f1_read_config32(0x44 + (node_id << 3));
+       f1_write_config32(0x44 + (node_id << 3),limit + (carry_over << 2));
+       dev = __f1_dev[node_id];
+       base  = pci_read_config32(dev, 0x40 + (node_id << 3));
+       basek  = (base & 0xffff0000) >> 2;
+       if(basek == hole_startk) {
+               //don't need set memhole here, because hole off set will be 0, overflow
+               //so need to change base reg instead, new basek will be 4*1024*1024
+               base &= 0x0000ffff;
+               base |= (4*1024*1024)<<2;
+               f1_write_config32(0x40 + (node_id<<3), base);
+       }
+       else if (dev)
+       {
+               hoist = /* hole start address */
+                       ((hole_startk << 10) & 0xff000000) +
+                       /* hole address to memory controller address */
+                       (((basek + carry_over) >> 6) & 0x0000ff00) +
+                       /* enable */
+                       1;
+
+               pci_write_config32(dev, 0xf0, hoist);
+       }
+
+       return carry_over;
+}
+#endif
+
+#if CONFIG_WRITE_HIGH_TABLES==1
+#define HIGH_TABLES_SIZE 64    // maximum size of high tables in KB
+extern uint64_t high_tables_base, high_tables_size;
+#if CONFIG_GFXUMA == 1
+extern uint64_t uma_memory_base, uma_memory_size;
+#endif
+#endif
+
+static void amdk8_domain_set_resources(device_t dev)
+{
+#if CONFIG_PCI_64BIT_PREF_MEM == 1
+       struct resource *io, *mem1, *mem2;
+       struct resource *res;
+#endif
        unsigned long mmio_basek;
-       uint32_t pci_tolm;
+       u32 pci_tolm;
        int i, idx;
+#if CONFIG_HW_MEM_HOLE_SIZEK != 0
+       struct hw_mem_hole_info mem_hole;
+       u32 reset_memhole = 1;
+#endif
+
+#if 0
+       /* Place the IO devices somewhere safe */
+       io = find_resource(dev, 0);
+       io->base = DEVICE_IO_START;
+#endif
+#if CONFIG_PCI_64BIT_PREF_MEM == 1
+       /* Now reallocate the pci resources memory with the
+        * highest addresses I can manage.
+        */
+       mem1 = find_resource(dev, 1);
+       mem2 = find_resource(dev, 2);
 
-       pci_tolm = find_pci_tolm(&dev->link[0]);
+#if 1
+       printk(BIOS_DEBUG, "base1: 0x%08Lx limit1: 0x%08Lx size: 0x%08Lx align: %d\n",
+               mem1->base, mem1->limit, mem1->size, mem1->align);
+       printk(BIOS_DEBUG, "base2: 0x%08Lx limit2: 0x%08Lx size: 0x%08Lx align: %d\n",
+               mem2->base, mem2->limit, mem2->size, mem2->align);
+#endif
+
+       /* See if both resources have roughly the same limits */
+       if (((mem1->limit <= 0xffffffff) && (mem2->limit <= 0xffffffff)) ||
+               ((mem1->limit > 0xffffffff) && (mem2->limit > 0xffffffff)))
+       {
+               /* If so place the one with the most stringent alignment first
+                */
+               if (mem2->align > mem1->align) {
+                       struct resource *tmp;
+                       tmp = mem1;
+                       mem1 = mem2;
+                       mem2 = tmp;
+               }
+               /* Now place the memory as high up as it will go */
+               mem2->base = resource_max(mem2);
+               mem1->limit = mem2->base - 1;
+               mem1->base = resource_max(mem1);
+       }
+       else {
+               /* Place the resources as high up as they will go */
+               mem2->base = resource_max(mem2);
+               mem1->base = resource_max(mem1);
+       }
 
-#warning "FIXME handle interleaved nodes"
+#if 1
+       printk(BIOS_DEBUG, "base1: 0x%08Lx limit1: 0x%08Lx size: 0x%08Lx align: %d\n",
+               mem1->base, mem1->limit, mem1->size, mem1->align);
+       printk(BIOS_DEBUG, "base2: 0x%08Lx limit2: 0x%08Lx size: 0x%08Lx align: %d\n",
+               mem2->base, mem2->limit, mem2->size, mem2->align);
+#endif
+
+       for(res = dev->resource_list; res; res = res->next)
+       {
+               res->flags |= IORESOURCE_ASSIGNED;
+               res->flags |= IORESOURCE_STORED;
+               report_resource_stored(dev, res, "");
+       }
+#endif
+
+       pci_tolm = my_find_pci_tolm(dev->link_list);
+
+       // FIXME handle interleaved nodes. If you fix this here, please fix
+       // amdfam10, too.
        mmio_basek = pci_tolm >> 10;
        /* Round mmio_basek to something the processor can support */
        mmio_basek &= ~((1 << 6) -1);
 
-#if 1
-#warning "FIXME improve mtrr.c so we don't use up all of the mtrrs with a 64M MMIO hole"
-       /* Round the mmio hold to 64M */
+       // FIXME improve mtrr.c so we don't use up all of the mtrrs with a 64M
+       // MMIO hole. If you fix this here, please fix amdfam10, too.
+       /* Round the mmio hole to 64M */
        mmio_basek &= ~((64*1024) - 1);
+
+#if CONFIG_HW_MEM_HOLE_SIZEK != 0
+       /* if the hw mem hole is already set in raminit stage, here we will compare mmio_basek and hole_basek
+        * if mmio_basek is bigger that hole_basek and will use hole_basek as mmio_basek and we don't need to reset hole.
+        * otherwise We reset the hole to the mmio_basek
+        */
+       #if CONFIG_K8_REV_F_SUPPORT == 0
+               if (!is_cpu_pre_e0()) {
+       #endif
+
+               mem_hole = get_hw_mem_hole_info();
+
+               if ((mem_hole.node_id !=  -1) && (mmio_basek > mem_hole.hole_startk)) { //We will use hole_basek as mmio_basek, and we don't need to reset hole anymore
+                       mmio_basek = mem_hole.hole_startk;
+                       reset_memhole = 0;
+               }
+
+               //mmio_basek = 3*1024*1024; // for debug to meet boundary
+
+               if(reset_memhole) {
+                       if(mem_hole.node_id!=-1) { // We need to select CONFIG_HW_MEM_HOLE_SIZEK for raminit, it can not make hole_startk to some basek too....!
+                              // We need to reset our Mem Hole, because We want more big HOLE than we already set
+                              //Before that We need to disable mem hole at first, becase memhole could already be set on i+1 instead
+                               disable_hoist_memory(mem_hole.hole_startk, mem_hole.node_id);
+                       }
+
+               #if CONFIG_HW_MEM_HOLE_SIZE_AUTO_INC == 1
+                       //We need to double check if the mmio_basek is valid for hole setting, if it is equal to basek, we need to decrease it some
+                       u32 basek_pri;
+                       for (i = 0; i < fx_devs; i++) {
+                               u32 base;
+                               u32 basek;
+                               base  = f1_read_config32(0x40 + (i << 3));
+                               if ((base & ((1<<1)|(1<<0))) != ((1<<1)|(1<<0))) {
+                                       continue;
+                               }
+
+                               basek = (base & 0xffff0000) >> 2;
+                               if(mmio_basek == basek) {
+                                       mmio_basek -= (basek - basek_pri)>>1; // increase mem hole size to make sure it is on middle of pri node
+                                       break;
+                               }
+                               basek_pri = basek;
+                       }
+               #endif
+               }
+
+#if CONFIG_K8_REV_F_SUPPORT == 0
+       } // is_cpu_pre_e0
+#endif
+
 #endif
 
-       idx = 10;
-       for (i = 0; i < 8; i++) {
-               uint32_t base, limit;
-               unsigned basek, limitk, sizek;
+       idx = 0x10;
+       for(i = 0; i < fx_devs; i++) {
+               u32 base, limit;
+               u32 basek, limitk, sizek;
                base  = f1_read_config32(0x40 + (i << 3));
                limit = f1_read_config32(0x44 + (i << 3));
                if ((base & ((1<<1)|(1<<0))) != ((1<<1)|(1<<0))) {
@@ -662,20 +980,53 @@ static void pci_domain_set_resources(device_t dev)
 
                /* see if we need a hole from 0xa0000 to 0xbffff */
                if ((basek < ((8*64)+(8*16))) && (sizek > ((8*64)+(16*16)))) {
-                       ram_resource(dev, idx++, basek, ((8*64)+(8*16)) - basek);
+                       ram_resource(dev, (idx | i), basek, ((8*64)+(8*16)) - basek);
+                       idx += 0x10;
                        basek = (8*64)+(16*16);
                        sizek = limitk - ((8*64)+(16*16));
-                       
+
                }
 
-               
+
+#if CONFIG_GFXUMA == 1
+               printk(BIOS_DEBUG, "node %d : uma_memory_base/1024=0x%08llx, mmio_basek=0x%08lx, basek=0x%08x, limitk=0x%08x\n", i, uma_memory_base >> 10, mmio_basek, basek, limitk);
+               if ((uma_memory_base >> 10) < mmio_basek)
+                       printk(BIOS_ALERT, "node %d: UMA memory starts below mmio_basek\n", i);
+#else
+//             printk(BIOS_DEBUG, "node %d : mmio_basek=%08x, basek=%08x, limitk=%08x\n", i, mmio_basek, basek, limitk); //yhlu
+#endif
+
                /* See if I need to split the region to accomodate pci memory space */
-               if ((basek < mmio_basek) && (limitk > mmio_basek)) {
-                       if (basek < mmio_basek) {
+               if ( (basek < 4*1024*1024 ) && (limitk > mmio_basek) ) {
+                       if (basek <= mmio_basek) {
                                unsigned pre_sizek;
                                pre_sizek = mmio_basek - basek;
-                               ram_resource(dev, idx++, basek, pre_sizek);
-                               sizek -= pre_sizek;
+                               if(pre_sizek>0) {
+                                       ram_resource(dev, (idx | i), basek, pre_sizek);
+                                       idx += 0x10;
+                                       sizek -= pre_sizek;
+#if CONFIG_WRITE_HIGH_TABLES==1
+                                       if (high_tables_base==0) {
+                                       /* Leave some space for ACPI, PIRQ and MP tables */
+#if CONFIG_GFXUMA == 1
+                                               high_tables_base = uma_memory_base - (HIGH_TABLES_SIZE * 1024);
+#else
+                                               high_tables_base = (mmio_basek - HIGH_TABLES_SIZE) * 1024;
+#endif
+                                               high_tables_size = HIGH_TABLES_SIZE * 1024;
+                                               printk(BIOS_DEBUG, " split: %dK table at =%08llx\n", HIGH_TABLES_SIZE,
+                                                            high_tables_base);
+                                       }
+#endif
+                               }
+                               #if CONFIG_HW_MEM_HOLE_SIZEK != 0
+                               if(reset_memhole)
+                                       #if CONFIG_K8_REV_F_SUPPORT == 0
+                                       if(!is_cpu_pre_e0() )
+                                       #endif
+                                                sizek += hoist_memory(mmio_basek,i);
+                               #endif
+
                                basek = mmio_basek;
                        }
                        if ((basek + sizek) <= 4*1024*1024) {
@@ -686,95 +1037,295 @@ static void pci_domain_set_resources(device_t dev)
                                sizek -= (4*1024*1024 - mmio_basek);
                        }
                }
-               ram_resource(dev, idx++, basek, sizek);
+               /* If sizek == 0, it was split at mmio_basek without a hole.
+                * Don't create an empty ram_resource.
+                */
+               if (sizek)
+                       ram_resource(dev, (idx | i), basek, sizek);
+               idx += 0x10;
+#if CONFIG_WRITE_HIGH_TABLES==1
+               printk(BIOS_DEBUG, "%d: mmio_basek=%08lx, basek=%08x, limitk=%08x\n",
+                            i, mmio_basek, basek, limitk);
+               if (high_tables_base==0) {
+               /* Leave some space for ACPI, PIRQ and MP tables */
+#if CONFIG_GFXUMA == 1
+                       high_tables_base = uma_memory_base - (HIGH_TABLES_SIZE * 1024);
+#else
+                       high_tables_base = (limitk - HIGH_TABLES_SIZE) * 1024;
+#endif
+                       high_tables_size = HIGH_TABLES_SIZE * 1024;
+               }
+#endif
        }
-       assign_resources(&dev->link[0]);
+       assign_resources(dev->link_list);
+
 }
 
-static unsigned int pci_domain_scan_bus(device_t dev, unsigned int max)
+static u32 amdk8_domain_scan_bus(device_t dev, u32 max)
 {
-       unsigned reg;
+       u32 reg;
+       int i;
        /* Unmap all of the HT chains */
-       for (reg = 0xe0; reg <= 0xec; reg += 4) {
+       for(reg = 0xe0; reg <= 0xec; reg += 4) {
                f1_write_config32(reg, 0);
        }
-       max = pci_scan_bus(&dev->link[0], PCI_DEVFN(0x18, 0), 0xff, max);
+       max = pci_scan_bus(dev->link_list, PCI_DEVFN(0x18, 0), 0xff, max);
+
+       /* Tune the hypertransport transaction for best performance.
+        * Including enabling relaxed ordering if it is safe.
+        */
+       get_fx_devs();
+       for(i = 0; i < fx_devs; i++) {
+               device_t f0_dev;
+               f0_dev = __f0_dev[i];
+               if (f0_dev && f0_dev->enabled) {
+                       u32 httc;
+                       httc = pci_read_config32(f0_dev, HT_TRANSACTION_CONTROL);
+                       httc &= ~HTTC_RSP_PASS_PW;
+                       if (!dev->link_list->disable_relaxed_ordering) {
+                               httc |= HTTC_RSP_PASS_PW;
+                       }
+                       printk(BIOS_SPEW, "%s passpw: %s\n",
+                               dev_path(dev),
+                               (!dev->link_list->disable_relaxed_ordering)?
+                               "enabled":"disabled");
+                       pci_write_config32(f0_dev, HT_TRANSACTION_CONTROL, httc);
+               }
+       }
        return max;
 }
 
 static struct device_operations pci_domain_ops = {
-       .read_resources   = pci_domain_read_resources,
-       .set_resources    = pci_domain_set_resources,
-       .enable_resources = enable_childrens_resources,
-       .init             = 0,
-       .scan_bus         = pci_domain_scan_bus,
-       .ops_pci_bus      = &pci_cf8_conf1,
+       .read_resources   = amdk8_domain_read_resources,
+       .set_resources    = amdk8_domain_set_resources,
+       .enable_resources = NULL,
+       .init             = NULL,
+       .scan_bus         = amdk8_domain_scan_bus,
+       .ops_pci_bus      = &pci_cf8_conf1,
 };
 
-static unsigned int cpu_bus_scan(device_t dev, unsigned int max)
+static void add_more_links(device_t dev, unsigned total_links)
+{
+       struct bus *link, *last = NULL;
+       int link_num;
+
+       for (link = dev->link_list; link; link = link->next)
+               last = link;
+
+       if (last) {
+               int links = total_links - last->link_num;
+               link_num = last->link_num;
+               if (links > 0) {
+                       link = malloc(links*sizeof(*link));
+                       if (!link)
+                               die("Couldn't allocate more links!\n");
+                       memset(link, 0, links*sizeof(*link));
+                       last->next = link;
+               }
+       }
+       else {
+               link_num = -1;
+               link = malloc(total_links*sizeof(*link));
+               memset(link, 0, total_links*sizeof(*link));
+               dev->link_list = link;
+       }
+
+       for (link_num = link_num + 1; link_num < total_links; link_num++) {
+               link->link_num = link_num;
+               link->dev = dev;
+               link->next = link + 1;
+               last = link;
+               link = link->next;
+       }
+       last->next = NULL;
+}
+
+static u32 cpu_bus_scan(device_t dev, u32 max)
 {
        struct bus *cpu_bus;
-       int i;
-       int apic_id_offset = lapicid(); // bsp apicid
+       device_t dev_mc;
+       int bsp_apicid;
+       int i,j;
+       unsigned nb_cfg_54;
+       unsigned siblings;
+       int e0_later_single_core;
+       int disable_siblings;
+
+       nb_cfg_54 = 0;
+       sysconf.enabled_apic_ext_id = 0;
+       sysconf.lift_bsp_apicid = 0;
+       siblings = 0;
+
+       /* Find the bootstrap processors apicid */
+       bsp_apicid = lapicid();
+       sysconf.apicid_offset = bsp_apicid;
+
+       disable_siblings = !CONFIG_LOGICAL_CPUS;
+#if CONFIG_LOGICAL_CPUS == 1
+       get_option(&disable_siblings, "multi_core");
+#endif
 
-       /* Find which cpus are present */
-       cpu_bus = &dev->link[0];
-       for (i = 0; i < 8; i++) {
-               device_t dev, cpu;
-               struct device_path cpu_path;
+       // for pre_e0, nb_cfg_54 can not be set, (when you read it still is 0)
+       // How can I get the nb_cfg_54 of every node's nb_cfg_54 in bsp???
+       // and differ d0 and e0 single core
+       nb_cfg_54 = read_nb_cfg_54();
 
-               /* Find the cpu's memory controller */
-               dev = dev_find_slot(0, PCI_DEVFN(0x18 + i, 0));
+       dev_mc = dev_find_slot(0, PCI_DEVFN(0x18, 0));
+       if (!dev_mc) {
+               die("0:18.0 not found?");
+       }
 
-               /* Build the cpu device path */
-               cpu_path.type = DEVICE_PATH_APIC;
-               cpu_path.u.apic.apic_id = i;
+       sysconf.nodes = ((pci_read_config32(dev_mc, 0x60)>>4) & 7) + 1;
 
-               /* See if I can find the cpu */
-               cpu = find_dev_path(cpu_bus, &cpu_path);
 
-               /* Enable the cpu if I have the processor */
-               if (dev && dev->enabled) {
-                       if (!cpu) {
-                               cpu = alloc_dev(cpu_bus, &cpu_path);
+       if (pci_read_config32(dev_mc, 0x68) & (HTTC_APIC_EXT_ID|HTTC_APIC_EXT_BRD_CST))
+       {
+               sysconf.enabled_apic_ext_id = 1;
+               if(bsp_apicid == 0) {
+                       /* bsp apic id is not changed */
+                       sysconf.apicid_offset = CONFIG_APIC_ID_OFFSET;
+               } else
+               {
+                       sysconf.lift_bsp_apicid = 1;
+               }
+
+       }
+
+       /* Find which cpus are present */
+       cpu_bus = dev->link_list;
+       for(i = 0; i < sysconf.nodes; i++) {
+               device_t cpu_dev, cpu;
+               struct device_path cpu_path;
+
+               /* Find the cpu's pci device */
+               cpu_dev = dev_find_slot(0, PCI_DEVFN(0x18 + i, 3));
+               if (!cpu_dev) {
+                       /* If I am probing things in a weird order
+                        * ensure all of the cpu's pci devices are found.
+                        */
+                       int local_j;
+                       device_t dev_f0;
+                       for(local_j = 0; local_j <= 3; local_j++) {
+                               cpu_dev = pci_probe_dev(NULL, dev_mc->bus,
+                                       PCI_DEVFN(0x18 + i, local_j));
                        }
-                       if (cpu) {
-                               cpu->enabled = 1;
+                       /* Ok, We need to set the links for that device.
+                        * otherwise the device under it will not be scanned
+                        */
+                       dev_f0 = dev_find_slot(0, PCI_DEVFN(0x18+i,0));
+                       if(dev_f0) {
+                               add_more_links(dev_f0, 3);
                        }
                }
-               
-               /* Disable the cpu if I don't have the processor */
-               if (cpu && (!dev || !dev->enabled)) {
-                       cpu->enabled = 0;
+
+               e0_later_single_core = 0;
+               if (cpu_dev && cpu_dev->enabled) {
+                       j = pci_read_config32(cpu_dev, 0xe8);
+                       j = (j >> 12) & 3; // dev is func 3
+                       printk(BIOS_DEBUG, "  %s siblings=%d\n", dev_path(cpu_dev), j);
+
+                       if(nb_cfg_54) {
+                               // For e0 single core if nb_cfg_54 is set, apicid will be 0, 2, 4....
+                               //  ----> you can mixed single core e0 and dual core e0 at any sequence
+                               // That is the typical case
+
+                               if(j == 0 ){
+                                      #if CONFIG_K8_REV_F_SUPPORT == 0
+                                       e0_later_single_core = is_e0_later_in_bsp(i);  // single core
+                                      #else
+                                       e0_later_single_core = is_cpu_f0_in_bsp(i);  // We can read cpuid(1) from Func3
+                                      #endif
+                               } else {
+                                      e0_later_single_core = 0;
+                               }
+                               if(e0_later_single_core) {
+                                       printk(BIOS_DEBUG, "\tFound Rev E or Rev F later single core\n");
+
+                                       j=1;
+                               }
+
+                               if(siblings > j ) {
+                               }
+                               else {
+                                       siblings = j;
+                               }
+                       } else {
+                               siblings = j;
+                       }
                }
-               
-               /* Report what I have done */
-               if (cpu) {
-                        if(cpu->path.u.apic.apic_id<apic_id_offset) {
-                                cpu->path.u.apic.apic_id += apic_id_offset;
-                        }  
-                       printk_debug("CPU: %s %s\n", dev_path(cpu),
-                                    cpu->enabled?"enabled":"disabled");
+
+               u32 jj;
+               if(e0_later_single_core || disable_siblings) {
+                       jj = 0;
+               } else
+               {
+                       jj = siblings;
                }
+#if 0
+               jj = 0; // if create cpu core1 path in amd_siblings by core0
+#endif
+
+               for (j = 0; j <=jj; j++ ) {
+
+                       /* Build the cpu device path */
+                       cpu_path.type = DEVICE_PATH_APIC;
+                       cpu_path.apic.apic_id = i * (nb_cfg_54?(siblings+1):1) + j * (nb_cfg_54?1:8);
+
+                       /* See if I can find the cpu */
+                       cpu = find_dev_path(cpu_bus, &cpu_path);
+
+                       /* Enable the cpu if I have the processor */
+                       if (cpu_dev && cpu_dev->enabled) {
+                               if (!cpu) {
+                                       cpu = alloc_dev(cpu_bus, &cpu_path);
+                               }
+                               if (cpu) {
+                                       cpu->enabled = 1;
+                               }
+                       }
+
+                       /* Disable the cpu if I don't have the processor */
+                       if (cpu && (!cpu_dev || !cpu_dev->enabled)) {
+                               cpu->enabled = 0;
+                       }
+
+                       /* Report what I have done */
+                       if (cpu) {
+                               cpu->path.apic.node_id = i;
+                               cpu->path.apic.core_id = j;
+                               if(sysconf.enabled_apic_ext_id) {
+                                       if(sysconf.lift_bsp_apicid) {
+                                               cpu->path.apic.apic_id += sysconf.apicid_offset;
+                                       } else
+                                       {
+                                               if (cpu->path.apic.apic_id != 0)
+                                                       cpu->path.apic.apic_id += sysconf.apicid_offset;
+                                       }
+                               }
+                               printk(BIOS_DEBUG, "CPU: %s %s\n",
+                                       dev_path(cpu), cpu->enabled?"enabled":"disabled");
+                       }
+
+               } //j
        }
        return max;
 }
 
 static void cpu_bus_init(device_t dev)
 {
-       initialize_cpus(&dev->link[0]);
+       initialize_cpus(dev->link_list);
 }
 
-static void cpu_bus_noop(device_t dev) 
+static void cpu_bus_noop(device_t dev)
 {
 }
 
 static struct device_operations cpu_bus_ops = {
-       .read_resources   = cpu_bus_noop,
-       .set_resources    = cpu_bus_noop,
+       .read_resources   = cpu_bus_noop,
+       .set_resources    = cpu_bus_noop,
        .enable_resources = cpu_bus_noop,
-       .init             = cpu_bus_init,
-       .scan_bus         = cpu_bus_scan,
+       .init             = cpu_bus_init,
+       .scan_bus         = cpu_bus_scan,
 };
 
 static void root_complex_enable_dev(struct device *dev)