static ssize_t node_read_distance(struct device *dev,
struct device_attribute *attr, char * buf)
{
int nid = dev->id;
int len = 0;
int i;
/*
* buf is currently PAGE_SIZE in length and each node needs 4 chars
* at the most (distance + space or newline).
*/
BUILD_BUG_ON(MAX_NUMNODES * 4 > PAGE_SIZE);
for_each_online_node(i)
len += sprintf(buf + len, "%s%d", i ? " " : "", node_distance(nid, i));
len += sprintf(buf + len, "\n");
return len;
}
/**
* acpi_map_pxm_to_online_node - Map proximity ID to online node
* @pxm: ACPI proximity ID
*
* This is similar to acpi_map_pxm_to_node(), but always returns an online
* node. When the mapped node from a given proximity ID is offline, it
* looks up the node distance table and returns the nearest online node.
*
* ACPI device drivers, which are called after the NUMA initialization has
* completed in the kernel, can call this interface to obtain their device
* NUMA topology from ACPI tables. Such drivers do not have to deal with
* offline nodes. A node may be offline when a device proximity ID is
* unique, SRAT memory entry does not exist, or NUMA is disabled, ex.
* "numa=off" on x86.
*/
int acpi_map_pxm_to_online_node(int pxm)
{
int node, n, dist, min_dist;
node = acpi_map_pxm_to_node(pxm);
if (node == NUMA_NO_NODE)
node = 0;
if (!node_online(node)) {
min_dist = INT_MAX;
for_each_online_node(n) {
dist = node_distance(node, n);
if (dist < min_dist) {
min_dist = dist;
node = n;
}
}
}
void quick_core(vector<node> s, node a, node b, vector<node>& con)
{//位于a->b的左边区域的点集s，返回该区域中的凸包顶点，存储于数组con中
//递归终止条件
if(s.empty())
return;
//找出区域s中距离向量a->b所在直线距离最远的点far_pos
vector<node>::iterator far_pos;
double dist(0);
for(vector<node>::iterator it = s.begin(); it != s.end(); ++ it){
//点it到向量a->b所在直线的距离为a->b和a->it的叉积除以a->b的长度
double d = cross(vec(a, b), vec(a, *it)) / node_distance(a, b);
if(dist < d){
//若点it到a->b所在直线的距离d比dist大则更新dist
//并用far_pos记下当前点i是距离最远的点
far_pos = it;
dist = d;
}
}
//将far_pos点加入凸包
con.push_back(*far_pos);
//在s中删去far_pos，但需要保留一个备份far
node far(*far_pos);
s.erase(far_pos);
vector<node> lt, rt;
for(vector<node>::iterator it = s.begin(); it != s.end(); ++ it){
//遍历区域s中每个点i
//计算a->far和a->it的叉积，若为正，则点it在向量a->far的左边
if(cross(vec(a, far), vec(a, *it)) > 0)
lt.push_back(*it);
//计算far->b和far->i的叉积，若为正，则点it在far->b的左边
if(cross(vec(far, b), vec(far, *it)) > 0)
rt.push_back(*it);
}
//继续递归
//求向量a->far左边的区域中的凸包顶点
quick_core(lt, a, far, con);
//求向量far->b左边的区域中的凸包顶点
quick_core(rt, far, b, con);
}
static int __init pcpu_cpu_distance(unsigned int from, unsigned int to)
{
return node_distance(early_cpu_to_node(from), early_cpu_to_node(to));
}
void __init acpi_numa_arch_fixup(void)
{
int i, j, node_from, node_to;
/* If there's no SRAT, fix the phys_id and mark node 0 online */
if (srat_num_cpus == 0) {
node_set_online(0);
node_cpuid[0].phys_id = hard_smp_processor_id();
return;
}
/*
* MCD - This can probably be dropped now. No need for pxm ID to node ID
* mapping with sparse node numbering iff MAX_PXM_DOMAINS <= MAX_NUMNODES.
*/
nodes_clear(node_online_map);
for (i = 0; i < MAX_PXM_DOMAINS; i++) {
if (pxm_bit_test(i)) {
int nid = acpi_map_pxm_to_node(i);
node_set_online(nid);
}
}
/* set logical node id in memory chunk structure */
for (i = 0; i < num_node_memblks; i++)
node_memblk[i].nid = pxm_to_node(node_memblk[i].nid);
/* assign memory bank numbers for each chunk on each node */
for_each_online_node(i) {
int bank;
bank = 0;
for (j = 0; j < num_node_memblks; j++)
if (node_memblk[j].nid == i)
node_memblk[j].bank = bank++;
}
/* set logical node id in cpu structure */
for_each_possible_early_cpu(i)
node_cpuid[i].nid = pxm_to_node(node_cpuid[i].nid);
printk(KERN_INFO "Number of logical nodes in system = %d\n",
num_online_nodes());
printk(KERN_INFO "Number of memory chunks in system = %d\n",
num_node_memblks);
if (!slit_table)
return;
memset(numa_slit, -1, sizeof(numa_slit));
for (i = 0; i < slit_table->locality_count; i++) {
if (!pxm_bit_test(i))
continue;
node_from = pxm_to_node(i);
for (j = 0; j < slit_table->locality_count; j++) {
if (!pxm_bit_test(j))
continue;
node_to = pxm_to_node(j);
node_distance(node_from, node_to) =
slit_table->entry[i * slit_table->locality_count + j];
}
}
#ifdef SLIT_DEBUG
printk("ACPI 2.0 SLIT locality table:\n");
for_each_online_node(i) {
for_each_online_node(j)
printk("%03d ", node_distance(i, j));
printk("\n");
}
#endif
}
void __init
acpi_numa_arch_fixup (void)
{
int i, j, node_from, node_to;
/* If there's no SRAT, fix the phys_id and mark node 0 online */
if (srat_num_cpus == 0) {
node_set_online(0);
node_cpuid[0].phys_id = hard_smp_processor_id();
return;
}
/* calculate total number of nodes in system from PXM bitmap */
numnodes = 0; /* init total nodes in system */
memset(pxm_to_nid_map, -1, sizeof(pxm_to_nid_map));
memset(nid_to_pxm_map, -1, sizeof(nid_to_pxm_map));
for (i = 0; i < MAX_PXM_DOMAINS; i++) {
if (pxm_bit_test(i)) {
pxm_to_nid_map[i] = numnodes;
node_set_online(numnodes);
nid_to_pxm_map[numnodes++] = i;
}
}
/* set logical node id in memory chunk structure */
for (i = 0; i < num_node_memblks; i++)
node_memblk[i].nid = pxm_to_nid_map[node_memblk[i].nid];
/* assign memory bank numbers for each chunk on each node */
for (i = 0; i < numnodes; i++) {
int bank;
bank = 0;
for (j = 0; j < num_node_memblks; j++)
if (node_memblk[j].nid == i)
node_memblk[j].bank = bank++;
}
/* set logical node id in cpu structure */
for (i = 0; i < srat_num_cpus; i++)
node_cpuid[i].nid = pxm_to_nid_map[node_cpuid[i].nid];
printk(KERN_INFO "Number of logical nodes in system = %d\n", numnodes);
printk(KERN_INFO "Number of memory chunks in system = %d\n", num_node_memblks);
if (!slit_table) return;
memset(numa_slit, -1, sizeof(numa_slit));
for (i=0; i<slit_table->localities; i++) {
if (!pxm_bit_test(i))
continue;
node_from = pxm_to_nid_map[i];
for (j=0; j<slit_table->localities; j++) {
if (!pxm_bit_test(j))
continue;
node_to = pxm_to_nid_map[j];
node_distance(node_from, node_to) =
slit_table->entry[i*slit_table->localities + j];
}
}
#ifdef SLIT_DEBUG
printk("ACPI 2.0 SLIT locality table:\n");
for (i = 0; i < numnodes; i++) {
for (j = 0; j < numnodes; j++)
printk("%03d ", node_distance(i,j));
printk("\n");
}
#endif
}
static int sn_topology_show(struct seq_file *s, void *d)
{
int sz;
int pt;
int e = 0;
int i;
int j;
const char *slabname;
int ordinal;
cpumask_t cpumask;
char slice;
struct cpuinfo_ia64 *c;
struct sn_hwperf_port_info *ptdata;
struct sn_hwperf_object_info *p;
struct sn_hwperf_object_info *obj = d; /* this object */
struct sn_hwperf_object_info *objs = s->private; /* all objects */
int rack, bay, slot, slab;
u8 shubtype;
u8 system_size;
u8 sharing_size;
u8 partid;
u8 coher;
u8 nasid_shift;
u8 region_size;
u16 nasid_mask;
int nasid_msb;
int pci_bus_ordinal = 0;
if (obj == objs) {
seq_printf(s, "# sn_topology version 2\n");
seq_printf(s, "# objtype ordinal location partition"
" [attribute value [, ...]]\n");
if (ia64_sn_get_sn_info(0,
&shubtype, &nasid_mask, &nasid_shift, &system_size,
&sharing_size, &partid, &coher, ®ion_size))
BUG();
for (nasid_msb=63; nasid_msb > 0; nasid_msb--) {
if (((u64)nasid_mask << nasid_shift) & (1ULL << nasid_msb))
break;
}
seq_printf(s, "partition %u %s local "
"shubtype %s, "
"nasid_mask 0x%016lx, "
"nasid_bits %d:%d, "
"system_size %d, "
"sharing_size %d, "
"coherency_domain %d, "
"region_size %d\n",
partid, system_utsname.nodename,
shubtype ? "shub2" : "shub1",
(u64)nasid_mask << nasid_shift, nasid_msb, nasid_shift,
system_size, sharing_size, coher, region_size);
}
if (SN_HWPERF_FOREIGN(obj)) {
/* private in another partition: not interesting */
return 0;
}
for (i = 0; i < SN_HWPERF_MAXSTRING && obj->name[i]; i++) {
if (obj->name[i] == ' ')
obj->name[i] = '_';
}
slabname = sn_hwperf_get_slabname(obj, objs, &ordinal);
seq_printf(s, "%s %d %s %s asic %s", slabname, ordinal, obj->location,
obj->sn_hwp_this_part ? "local" : "shared", obj->name);
if (!SN_HWPERF_IS_NODE(obj) && !SN_HWPERF_IS_IONODE(obj))
seq_putc(s, '\n');
else {
seq_printf(s, ", nasid 0x%x", cnodeid_to_nasid(ordinal));
for (i=0; i < numionodes; i++) {
seq_printf(s, i ? ":%d" : ", dist %d",
node_distance(ordinal, i));
}
seq_putc(s, '\n');
/*
* CPUs on this node, if any
*/
cpumask = node_to_cpumask(ordinal);
for_each_online_cpu(i) {
if (cpu_isset(i, cpumask)) {
slice = 'a' + cpuid_to_slice(i);
c = cpu_data(i);
seq_printf(s, "cpu %d %s%c local"
" freq %luMHz, arch ia64",
i, obj->location, slice,
c->proc_freq / 1000000);
for_each_online_cpu(j) {
seq_printf(s, j ? ":%d" : ", dist %d",
node_distance(
cpuid_to_cnodeid(i),
cpuid_to_cnodeid(j)));
}
seq_putc(s, '\n');
}
}
/*
* PCI busses attached to this node, if any
*/
if (sn_hwperf_location_to_bpos(obj->location,
&rack, &bay, &slot, &slab)) {
/* export pci bus info */
print_pci_topology(s, obj, &pci_bus_ordinal,
rack, bay, slot, slab);
}
}
if (obj->ports) {
/*
* numalink ports
*/
sz = obj->ports * sizeof(struct sn_hwperf_port_info);
if ((ptdata = vmalloc(sz)) == NULL)
return -ENOMEM;
e = ia64_sn_hwperf_op(sn_hwperf_master_nasid,
SN_HWPERF_ENUM_PORTS, obj->id, sz,
(u64) ptdata, 0, 0, NULL);
if (e != SN_HWPERF_OP_OK)
return -EINVAL;
for (ordinal=0, p=objs; p != obj; p++) {
if (!SN_HWPERF_FOREIGN(p))
ordinal += p->ports;
}
for (pt = 0; pt < obj->ports; pt++) {
for (p = objs, i = 0; i < sn_hwperf_obj_cnt; i++, p++) {
if (ptdata[pt].conn_id == p->id) {
break;
}
}
seq_printf(s, "numalink %d %s-%d",
ordinal+pt, obj->location, ptdata[pt].port);
if (i >= sn_hwperf_obj_cnt) {
/* no connection */
seq_puts(s, " local endpoint disconnected"
", protocol unknown\n");
continue;
}
if (obj->sn_hwp_this_part && p->sn_hwp_this_part)
/* both ends local to this partition */
seq_puts(s, " local");
else if (!obj->sn_hwp_this_part && !p->sn_hwp_this_part)
/* both ends of the link in foreign partiton */
seq_puts(s, " foreign");
else
/* link straddles a partition */
seq_puts(s, " shared");
/*
* Unlikely, but strictly should query the LLP config
* registers because an NL4R can be configured to run
* NL3 protocol, even when not talking to an NL3 router.
* Ditto for node-node.
*/
seq_printf(s, " endpoint %s-%d, protocol %s\n",
p->location, ptdata[pt].conn_port,
(SN_HWPERF_IS_NL3ROUTER(obj) ||
SN_HWPERF_IS_NL3ROUTER(p)) ? "LLP3" : "LLP4");
}
vfree(ptdata);
}
return 0;
}
void __init acpi_numa_arch_fixup(void)
{
int i, j, node_from, node_to;
if (srat_num_cpus == 0) {
node_set_online(0);
node_cpuid[0].phys_id = hard_smp_processor_id();
return;
}
nodes_clear(node_online_map);
for (i = 0; i < MAX_PXM_DOMAINS; i++) {
if (pxm_bit_test(i)) {
int nid = acpi_map_pxm_to_node(i);
node_set_online(nid);
}
}
for (i = 0; i < num_node_memblks; i++)
node_memblk[i].nid = pxm_to_node(node_memblk[i].nid);
for_each_online_node(i) {
int bank;
bank = 0;
for (j = 0; j < num_node_memblks; j++)
if (node_memblk[j].nid == i)
node_memblk[j].bank = bank++;
}
for_each_possible_early_cpu(i)
node_cpuid[i].nid = pxm_to_node(node_cpuid[i].nid);
printk(KERN_INFO "Number of logical nodes in system = %d\n",
num_online_nodes());
printk(KERN_INFO "Number of memory chunks in system = %d\n",
num_node_memblks);
if (!slit_table) {
for (i = 0; i < MAX_NUMNODES; i++)
for (j = 0; j < MAX_NUMNODES; j++)
node_distance(i, j) = i == j ? LOCAL_DISTANCE :
REMOTE_DISTANCE;
return;
}
memset(numa_slit, -1, sizeof(numa_slit));
for (i = 0; i < slit_table->locality_count; i++) {
if (!pxm_bit_test(i))
continue;
node_from = pxm_to_node(i);
for (j = 0; j < slit_table->locality_count; j++) {
if (!pxm_bit_test(j))
continue;
node_to = pxm_to_node(j);
node_distance(node_from, node_to) =
slit_table->entry[i * slit_table->locality_count + j];
}
}
#ifdef SLIT_DEBUG
printk("ACPI 2.0 SLIT locality table:\n");
for_each_online_node(i) {
for_each_online_node(j)
printk("%03d ", node_distance(i, j));
printk("\n");
}
#endif
}
