@@ -11,6 +11,7 @@
#include <xen/sched.h>
#include <xen/sched-if.h>
#include <xen/softirq.h>
+#include <acpi/srat.h>
#include <asm/dom0_build.h>
#include <asm/hpet.h>
@@ -137,8 +137,6 @@ l1_pgentry_t __section(".bss.page_aligned") __aligned(PAGE_SIZE)
#define PTE_UPDATE_WITH_CMPXCHG
#endif
-paddr_t __read_mostly mem_hotplug;
-
/* Private domain structs for DOMID_XEN and DOMID_IO. */
struct domain *dom_xen, *dom_io, *dom_cow;
@@ -8,6 +8,7 @@
#include <xen/guest_access.h>
#include <xen/iocap.h>
#include <xen/serial.h>
+#include <acpi/srat.h>
#include <asm/current.h>
#include <asm/io_apic.h>
#include <asm/msi.h>
@@ -27,6 +27,7 @@
#include <xen/tmem_xen.h>
#include <xen/virtual_region.h>
#include <xen/watchdog.h>
+#include <acpi/srat.h>
#include <public/version.h>
#include <compat/platform.h>
#include <compat/xen.h>
@@ -33,6 +33,7 @@
#include <xen/serial.h>
#include <xen/numa.h>
#include <xen/cpu.h>
+#include <acpi/srat.h>
#include <asm/current.h>
#include <asm/mc146818rtc.h>
#include <asm/desc.h>
@@ -18,14 +18,12 @@
#include <xen/acpi.h>
#include <xen/numa.h>
#include <xen/pfn.h>
+#include <acpi/srat.h>
#include <asm/e820.h>
#include <asm/page.h>
-static struct acpi_table_slit *__read_mostly acpi_slit;
-
extern nodemask_t processor_nodes_parsed;
extern nodemask_t memory_nodes_parsed;
-
/*
* Keep BIOS's CPU2node information, should not be used for memory allocaion
*/
@@ -33,87 +31,6 @@ nodeid_t apicid_to_node[MAX_LOCAL_APIC] = {
[0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE
};
-struct pxm2node {
- unsigned int pxm;
- nodeid_t node;
-};
-static struct pxm2node __read_mostly pxm2node[MAX_NUMNODES] =
- { [0 ... MAX_NUMNODES - 1] = {.node = NUMA_NO_NODE} };
-
-static unsigned node_to_pxm(nodeid_t n);
-
-static __initdata DECLARE_BITMAP(memblk_hotplug, NR_NODE_MEMBLKS);
-
-static inline bool node_found(unsigned int idx, unsigned int pxm)
-{
- return ((pxm2node[idx].pxm == pxm) &&
- (pxm2node[idx].node != NUMA_NO_NODE));
-}
-
-static void reset_pxm2node(void)
-{
- unsigned int i;
-
- for (i = 0; i < ARRAY_SIZE(pxm2node); i++)
- pxm2node[i].node = NUMA_NO_NODE;
-}
-
-nodeid_t pxm_to_node(unsigned int pxm)
-{
- unsigned int i;
-
- if ((pxm < ARRAY_SIZE(pxm2node)) && node_found(pxm, pxm))
- return pxm2node[pxm].node;
-
- for (i = 0; i < ARRAY_SIZE(pxm2node); i++)
- if (node_found(i, pxm))
- return pxm2node[i].node;
-
- return NUMA_NO_NODE;
-}
-
-nodeid_t acpi_setup_node(unsigned int pxm)
-{
- nodeid_t node;
- unsigned int idx;
- static bool warned;
- static unsigned int nodes_found;
-
- BUILD_BUG_ON(MAX_NUMNODES >= NUMA_NO_NODE);
-
- if (pxm < ARRAY_SIZE(pxm2node)) {
- if (node_found(pxm, pxm))
- return pxm2node[pxm].node;
-
- /* Try to maintain indexing of pxm2node by pxm */
- if (pxm2node[pxm].node == NUMA_NO_NODE) {
- idx = pxm;
- goto finish;
- }
- }
-
- for (idx = 0; idx < ARRAY_SIZE(pxm2node); idx++)
- if (pxm2node[idx].node == NUMA_NO_NODE)
- goto finish;
-
- if (!warned) {
- printk(KERN_WARNING "SRAT: Too many proximity domains (%#x)\n",
- pxm);
- warned = 1;
- }
-
- return NUMA_NO_NODE;
-
- finish:
- node = nodes_found++;
- if (node >= MAX_NUMNODES)
- return NUMA_NO_NODE;
- pxm2node[idx].pxm = pxm;
- pxm2node[idx].node = node;
-
- return node;
-}
-
void __init numa_failed(void)
{
int i;
@@ -125,48 +42,6 @@ void __init numa_failed(void)
mem_hotplug = 0;
}
-/*
- * A lot of BIOS fill in 10 (= no distance) everywhere. This messes
- * up the NUMA heuristics which wants the local node to have a smaller
- * distance than the others.
- * Do some quick checks here and only use the SLIT if it passes.
- */
-static int __init slit_valid(struct acpi_table_slit *slit)
-{
- int i, j;
- int d = slit->locality_count;
- for (i = 0; i < d; i++) {
- for (j = 0; j < d; j++) {
- uint8_t val = slit->entry[d*i + j];
- if (i == j) {
- if (val != LOCAL_DISTANCE)
- return 0;
- } else if (val <= LOCAL_DISTANCE)
- return 0;
- }
- }
- return 1;
-}
-
-/* Callback for SLIT parsing */
-void __init acpi_numa_slit_init(struct acpi_table_slit *slit)
-{
- unsigned long mfn;
- if (!slit_valid(slit)) {
- printk(KERN_INFO "ACPI: SLIT table looks invalid. "
- "Not used.\n");
- return;
- }
- mfn = alloc_boot_pages(PFN_UP(slit->header.length), 1);
- if (!mfn) {
- printk(KERN_ERR "ACPI: Unable to allocate memory for "
- "saving ACPI SLIT numa information.\n");
- return;
- }
- acpi_slit = mfn_to_virt(mfn);
- memcpy(acpi_slit, slit, slit->header.length);
-}
-
/* Callback for Proximity Domain -> x2APIC mapping */
void __init
acpi_numa_x2apic_affinity_init(const struct acpi_srat_x2apic_cpu_affinity *pa)
@@ -234,100 +109,6 @@ acpi_numa_processor_affinity_init(const struct acpi_srat_cpu_affinity *pa)
pxm, pa->apic_id, node);
}
-/* Callback for parsing of the Proximity Domain <-> Memory Area mappings */
-void __init
-acpi_numa_memory_affinity_init(const struct acpi_srat_mem_affinity *ma)
-{
- uint64_t start, end;
- unsigned pxm;
- nodeid_t node;
- int i;
- struct node *memblk;
-
- if (srat_disabled())
- return;
- if (ma->header.length != sizeof(struct acpi_srat_mem_affinity)) {
- numa_failed();
- return;
- }
- if (!(ma->flags & ACPI_SRAT_MEM_ENABLED))
- return;
-
- if (get_num_node_memblks() >= NR_NODE_MEMBLKS)
- {
- dprintk(XENLOG_WARNING,
- "Too many numa entry, try bigger NR_NODE_MEMBLKS \n");
- numa_failed();
- return;
- }
-
- start = ma->base_address;
- end = start + ma->length;
- pxm = ma->proximity_domain;
- if (srat_rev < 2)
- pxm &= 0xff;
- node = acpi_setup_node(pxm);
- if (node == NUMA_NO_NODE) {
- numa_failed();
- return;
- }
- /* It is fine to add this area to the nodes data it will be used later*/
- i = conflicting_memblks(start, end);
- if (i < 0)
- /* everything fine */;
- else if (get_memblk_nodeid(i) == node) {
- bool mismatch = !(ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE) !=
- !test_bit(i, memblk_hotplug);
-
- memblk = get_node_memblk_range(i);
-
- printk("%sSRAT: PXM %u (%"PRIx64"-%"PRIx64") overlaps with itself (%"PRIx64"-%"PRIx64")\n",
- mismatch ? KERN_ERR : KERN_WARNING, pxm, start, end,
- memblk->start, memblk->end);
- if (mismatch) {
- numa_failed();
- return;
- }
- } else {
- memblk = get_node_memblk_range(i);
-
- printk(KERN_ERR
- "SRAT: PXM %u (%"PRIx64"-%"PRIx64") overlaps with PXM %u (%"PRIx64"-%"PRIx64")\n",
- pxm, start, end, node_to_pxm(get_memblk_nodeid(i)),
- memblk->start, memblk->end);
- numa_failed();
- return;
- }
- if (!(ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE)) {
- struct node *nd = get_numa_node(node);
-
- if (!node_test_and_set(node, memory_nodes_parsed)) {
- nd->start = start;
- nd->end = end;
- } else {
- if (start < nd->start)
- nd->start = start;
- if (nd->end < end)
- nd->end = end;
- }
- }
- printk(KERN_INFO "SRAT: Node %u PXM %u %"PRIx64"-%"PRIx64"%s\n",
- node, pxm, start, end,
- ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE ? " (hotplug)" : "");
-
- if (numa_add_memblk(node, start, ma->length)) {
- printk(KERN_ERR "SRAT: node-id %u out of range\n", node);
- numa_failed();
- return;
- }
-
- if (ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE) {
- __set_bit(get_num_node_memblks(), memblk_hotplug);
- if (end > mem_hotplug)
- mem_hotplug = end;
- }
-}
-
/* Sanity check to catch more bad SRATs (they are amazingly common).
Make sure the PXMs cover all memory. */
int __init arch_sanitize_nodes_memory(void)
@@ -427,35 +208,6 @@ void __init srat_parse_regions(uint64_t addr)
pfn_pdx_hole_setup(mask >> PAGE_SHIFT);
}
-static unsigned node_to_pxm(nodeid_t n)
-{
- unsigned i;
-
- if ((n < ARRAY_SIZE(pxm2node)) && (pxm2node[n].node == n))
- return pxm2node[n].pxm;
- for (i = 0; i < ARRAY_SIZE(pxm2node); i++)
- if (pxm2node[i].node == n)
- return pxm2node[i].pxm;
- return 0;
-}
-
-static uint8_t acpi_node_distance(nodeid_t a, nodeid_t b)
-{
- unsigned index;
- uint8_t slit_val;
-
- if (!acpi_slit)
- return a == b ? LOCAL_DISTANCE : REMOTE_DISTANCE;
- index = acpi_slit->locality_count * node_to_pxm(a);
- slit_val = acpi_slit->entry[index + node_to_pxm(b)];
-
- /* ACPI defines 0xff as an unreachable node and 0-9 are undefined */
- if ((slit_val == 0xff) || (slit_val <= 9))
- return NUMA_NO_DISTANCE;
- else
- return slit_val;
-}
-
uint8_t __node_distance(nodeid_t a, nodeid_t b)
{
return acpi_node_distance(a, b);
@@ -27,6 +27,7 @@ asm(".file \"" __FILE__ "\"");
#include <xen/guest_access.h>
#include <xen/hypercall.h>
#include <xen/mem_access.h>
+#include <acpi/srat.h>
#include <asm/current.h>
#include <asm/asm_defns.h>
#include <asm/page.h>
@@ -4,6 +4,7 @@ subdir-$(CONFIG_X86) += apei
obj-bin-y += tables.init.o
obj-$(CONFIG_NUMA) += numa.o
+obj-$(CONFIG_NUMA) += srat.o
obj-y += osl.o
obj-$(CONFIG_HAS_CPUFREQ) += pmstat.o
new file mode 100644
@@ -0,0 +1,299 @@
+/*
+ * ACPI 3.0 based NUMA setup
+ * Copyright 2004 Andi Kleen, SuSE Labs.
+ *
+ * Reads the ACPI SRAT table to figure out what memory belongs to which CPUs.
+ *
+ * Called from acpi_numa_init while reading the SRAT and SLIT tables.
+ * Assumes all memory regions belonging to a single proximity domain
+ * are in one chunk. Holes between them will be included in the node.
+ *
+ * Adapted for Xen: Ryan Harper <ryanh@us.ibm.com>
+ */
+
+#include <xen/init.h>
+#include <xen/mm.h>
+#include <xen/inttypes.h>
+#include <xen/nodemask.h>
+#include <xen/acpi.h>
+#include <xen/numa.h>
+#include <xen/pfn.h>
+#include <acpi/srat.h>
+#include <asm/page.h>
+#include <asm/acpi.h>
+
+paddr_t __read_mostly mem_hotplug;
+extern nodemask_t memory_nodes_parsed;
+static struct acpi_table_slit __read_mostly *acpi_slit;
+static struct pxm2node __read_mostly pxm2node[MAX_NUMNODES] =
+ { [0 ... MAX_NUMNODES - 1] = {.node = NUMA_NO_NODE} };
+
+static __initdata DECLARE_BITMAP(memblk_hotplug, NR_NODE_MEMBLKS);
+
+static inline bool_t node_found(unsigned int idx, unsigned int pxm)
+{
+ return ( (pxm2node[idx].pxm == pxm) &&
+ (pxm2node[idx].node != NUMA_NO_NODE) );
+}
+
+void reset_pxm2node(void)
+{
+ unsigned int i;
+
+ for ( i = 0; i < ARRAY_SIZE(pxm2node); i++ )
+ pxm2node[i].node = NUMA_NO_NODE;
+}
+
+unsigned node_to_pxm(nodeid_t n)
+{
+ unsigned int i;
+
+ if ( (n < ARRAY_SIZE(pxm2node)) && (pxm2node[n].node == n) )
+ return pxm2node[n].pxm;
+
+ for ( i = 0; i < ARRAY_SIZE(pxm2node); i++ )
+ if ( pxm2node[i].node == n )
+ return pxm2node[i].pxm;
+
+ return 0;
+}
+
+nodeid_t pxm_to_node(unsigned int pxm)
+{
+ unsigned int i;
+
+ if ( (pxm < ARRAY_SIZE(pxm2node)) && node_found(pxm, pxm) )
+ return pxm2node[pxm].node;
+
+ for ( i = 0; i < ARRAY_SIZE(pxm2node); i++ )
+ if ( node_found(i, pxm) )
+ return pxm2node[i].node;
+
+ return NUMA_NO_NODE;
+}
+
+nodeid_t acpi_setup_node(unsigned int pxm)
+{
+ nodeid_t node;
+ unsigned int idx;
+ static bool_t warned;
+ static unsigned int nodes_found;
+
+ BUILD_BUG_ON(MAX_NUMNODES >= NUMA_NO_NODE);
+
+ if ( pxm < ARRAY_SIZE(pxm2node) )
+ {
+ if ( node_found(pxm, pxm) )
+ return pxm2node[pxm].node;
+
+ /* Try to maintain indexing of pxm2node by pxm */
+ if ( pxm2node[pxm].node == NUMA_NO_NODE )
+ {
+ idx = pxm;
+ goto finish;
+ }
+ }
+
+ for ( idx = 0; idx < ARRAY_SIZE(pxm2node); idx++ )
+ if ( pxm2node[idx].node == NUMA_NO_NODE )
+ goto finish;
+
+ if ( !warned )
+ {
+ printk(KERN_WARNING "SRAT: Too many proximity domains (%#x)\n", pxm);
+ warned = 1;
+ }
+
+ return NUMA_NO_NODE;
+
+ finish:
+ node = nodes_found++;
+ if ( node >= MAX_NUMNODES )
+ return NUMA_NO_NODE;
+ pxm2node[idx].pxm = pxm;
+ pxm2node[idx].node = node;
+
+ return node;
+}
+
+/*
+ * A lot of BIOS fill in 10 (= no distance) everywhere. This messes
+ * up the NUMA heuristics which wants the local node to have a smaller
+ * distance than the others.
+ * Do some quick checks here and only use the SLIT if it passes.
+ */
+static __init int slit_valid(struct acpi_table_slit *slit)
+{
+ int i, j;
+ int d = slit->locality_count;
+
+ for ( i = 0; i < d; i++ )
+ {
+ for ( j = 0; j < d; j++ )
+ {
+ uint8_t val = slit->entry[d * i + j];
+
+ if ( i == j )
+ {
+ if ( val != LOCAL_DISTANCE )
+ return 0;
+ } else if ( val <= LOCAL_DISTANCE )
+ return 0;
+ }
+ }
+
+ return 1;
+}
+
+/* Callback for SLIT parsing */
+void __init acpi_numa_slit_init(struct acpi_table_slit *slit)
+{
+ unsigned long mfn;
+
+ if ( !slit_valid(slit) )
+ {
+ printk(KERN_INFO "ACPI: SLIT table looks invalid. Not used.\n");
+ return;
+ }
+
+ mfn = alloc_boot_pages(PFN_UP(slit->header.length), 1);
+ if ( !mfn )
+ {
+ printk(KERN_ERR "ACPI: Unable to allocate memory for "
+ "saving ACPI SLIT numa information.\n");
+ return;
+ }
+ acpi_slit = mfn_to_virt(mfn);
+ memcpy(acpi_slit, slit, slit->header.length);
+}
+
+/* Callback for parsing of the Proximity Domain <-> Memory Area mappings */
+void __init
+acpi_numa_memory_affinity_init(const struct acpi_srat_mem_affinity *ma)
+{
+ uint64_t start, end;
+ unsigned pxm;
+ nodeid_t node;
+ int i;
+ struct node *memblk;
+
+ if ( srat_disabled() )
+ return;
+
+ if ( ma->header.length != sizeof(struct acpi_srat_mem_affinity) )
+ {
+ numa_failed();
+ return;
+ }
+
+ if ( !(ma->flags & ACPI_SRAT_MEM_ENABLED) )
+ return;
+
+ if ( get_num_node_memblks() >= NR_NODE_MEMBLKS )
+ {
+ dprintk(XENLOG_WARNING,
+ "Too many numa entry, try bigger NR_NODE_MEMBLKS \n");
+ numa_failed();
+ return;
+ }
+
+ start = ma->base_address;
+ end = start + ma->length;
+ pxm = ma->proximity_domain;
+ if ( srat_rev < 2 )
+ pxm &= 0xff;
+ node = acpi_setup_node(pxm);
+ if ( node == NUMA_NO_NODE )
+ {
+ numa_failed();
+ return;
+ }
+
+ /* It is fine to add this area to the nodes data it will be used later*/
+ i = conflicting_memblks(start, end);
+ if ( i < 0 )
+ /* everything fine */;
+ else if ( get_memblk_nodeid(i) == node )
+ {
+ bool_t mismatch = !(ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE) !=
+ !test_bit(i, memblk_hotplug);
+
+ memblk = get_node_memblk_range(i);
+
+ printk("%sSRAT: PXM %u (%"PRIx64"-%"PRIx64") overlaps with itself (%"PRIx64"-%"PRIx64")\n",
+ mismatch ? KERN_ERR : KERN_WARNING, pxm, start, end,
+ memblk->start, memblk->end);
+ if ( mismatch )
+ {
+ numa_failed();
+ return;
+ }
+ }
+ else
+ {
+ memblk = get_node_memblk_range(i);
+
+ printk(KERN_ERR
+ "SRAT: PXM %u (%"PRIx64"-%"PRIx64") overlaps with PXM %u (%"PRIx64"-%"PRIx64")\n",
+ pxm, start, end, node_to_pxm(get_memblk_nodeid(i)),
+ memblk->start, memblk->end);
+ numa_failed();
+ return;
+ }
+
+ if ( !(ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE) )
+ {
+ struct node *nd = get_numa_node(node);
+
+ if ( !node_test_and_set(node, memory_nodes_parsed) )
+ {
+ nd->start = start;
+ nd->end = end;
+ }
+ else
+ {
+ if ( start < nd->start )
+ nd->start = start;
+ if ( nd->end < end )
+ nd->end = end;
+ }
+ }
+ printk(KERN_INFO "SRAT: Node %u PXM %u %"PRIx64"-%"PRIx64"%s\n",
+ node, pxm, start, end,
+ ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE ? " (hotplug)" : "");
+
+ numa_add_memblk(node, start, ma->length);
+ if ( ma->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE )
+ {
+ __set_bit(get_num_node_memblks(), memblk_hotplug);
+ if ( end > mem_hotplug )
+ mem_hotplug = end;
+ }
+}
+
+uint8_t acpi_node_distance(nodeid_t a, nodeid_t b)
+{
+ unsigned index;
+ uint8_t slit_val;
+
+ if ( !acpi_slit )
+ return a == b ? LOCAL_DISTANCE : REMOTE_DISTANCE;
+
+ index = acpi_slit->locality_count * node_to_pxm(a);
+ slit_val = acpi_slit->entry[index + node_to_pxm(b)];
+
+ /* ACPI defines 0xff as an unreachable node and 0-9 are undefined */
+ if ( (slit_val == 0xff) || (slit_val <= 9) )
+ return NUMA_NO_DISTANCE;
+ else
+ return slit_val;
+}
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
@@ -30,6 +30,7 @@
#include <xen/pci.h>
#include <xen/pci_regs.h>
#include <xen/keyhandler.h>
+#include <acpi/srat.h>
#include <asm/msi.h>
#include <asm/irq.h>
#include <asm/hvm/vmx/vmx.h>
new file mode 100644
@@ -0,0 +1,24 @@
+#ifndef __XEN_SRAT_H__
+#define __XEN_SRAT_H__
+
+extern int srat_rev;
+struct pxm2node {
+ unsigned int pxm;
+ nodeid_t node;
+};
+
+extern nodeid_t pxm_to_node(unsigned pxm);
+extern nodeid_t acpi_setup_node(unsigned pxm);
+extern unsigned int node_to_pxm(nodeid_t n);
+extern uint8_t acpi_node_distance(nodeid_t a, nodeid_t b);
+extern void reset_pxm2node(void);
+#endif /* __XEN_SRAT_H__ */
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
@@ -405,7 +405,6 @@ static inline int get_page_and_type(struct page_info *page,
int check_descriptor(const struct domain *, struct desc_struct *d);
extern bool_t opt_allow_superpage;
-extern paddr_t mem_hotplug;
/******************************************************************************
* With shadow pagetables, the different kinds of address start
@@ -7,8 +7,6 @@
typedef uint8_t nodeid_t;
-extern int srat_rev;
-
extern nodeid_t cpu_to_node[NR_CPUS];
extern cpumask_t node_to_cpumask[];
@@ -17,8 +15,6 @@ extern cpumask_t node_to_cpumask[];
#define node_to_first_cpu(node) (__ffs(node_to_cpumask[node]))
#define node_to_cpumask(node) (node_to_cpumask[node])
-extern nodeid_t pxm_to_node(unsigned int pxm);
-
#define ZONE_ALIGN (1UL << (MAX_ORDER+PAGE_SHIFT))
extern void numa_add_cpu(int cpu);
@@ -63,6 +63,8 @@ TYPE_SAFE(unsigned long, mfn);
#undef mfn_t
#endif
+extern paddr_t mem_hotplug;
+
static inline mfn_t mfn_add(mfn_t mfn, unsigned long i)
{
return _mfn(mfn_x(mfn) + i);
@@ -25,7 +25,6 @@ extern int compute_memnode_shift(struct node *nodes, int numnodes,
extern void numa_init_array(void);
extern bool_t srat_disabled(void);
extern void numa_set_node(int cpu, nodeid_t node);
-extern nodeid_t acpi_setup_node(unsigned int pxm);
extern void srat_detect_node(int cpu);
extern void setup_node_bootmem(nodeid_t nodeid, paddr_t start, paddr_t end);
extern void init_cpu_to_node(void);