linux/arch/x86/kernel/srat_32.c
Yinghai Lu 9a73aa81ff x86: 32bit numa srat fix early_ioremap leak
on two node system (16g RAM) with numa config I got this crash:

get_memcfg_from_srat: assigning address to rsdp
RSD PTR  v0 [ACPIAM]
ACPI: Too big length in RSDT: 92
failed to get NUMA memory information from SRAT table
NUMA - single node, flat memory mode
Node: 0, start_pfn: 0, end_pfn: 153
 Setting physnode_map array to node 0 for pfns:
 0
...
Pid: 0, comm: swapper Not tainted 2.6.26-rc4 #4
 [<80b41289>] hlt_loop+0x0/0x3
 [<8011efa0>] ? alloc_remap+0x50/0x70
 [<8079e32e>] alloc_node_mem_map+0x5e/0xa0
 [<8012e77b>] ? printk+0x1b/0x20
 [<80b590f6>] free_area_init_node+0xc6/0x470
 [<80b588fc>] ? __alloc_bootmem_node+0x2c/0x50
 [<80b58ad8>] ? find_min_pfn_for_node+0x38/0x70
 [<8012e77b>] ? printk+0x1b/0x20
 [<80b597c4>] free_area_init_nodes+0x254/0x2d0
 [<80b544d7>] zone_sizes_init+0x97/0xa0
 [<80b48a03>] setup_arch+0x383/0x530
 [<8012e77b>] ? printk+0x1b/0x20
 [<80b41aa4>] start_kernel+0x64/0x350
 [<80b412d8>] i386_start_kernel+0x8/0x10
 =======================

this patch increases the acpi table limit to 32.
Also match early_ioremap() with early_iounmap().

Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-05-31 09:55:56 +02:00

374 lines
11 KiB
C

/*
* Some of the code in this file has been gleaned from the 64 bit
* discontigmem support code base.
*
* Copyright (C) 2002, IBM Corp.
*
* All rights reserved.
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
* NON INFRINGEMENT. See the GNU General Public License for more
* details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*
* Send feedback to Pat Gaughen <gone@us.ibm.com>
*/
#include <linux/mm.h>
#include <linux/bootmem.h>
#include <linux/mmzone.h>
#include <linux/acpi.h>
#include <linux/nodemask.h>
#include <asm/srat.h>
#include <asm/topology.h>
#include <asm/smp.h>
/*
* proximity macros and definitions
*/
#define NODE_ARRAY_INDEX(x) ((x) / 8) /* 8 bits/char */
#define NODE_ARRAY_OFFSET(x) ((x) % 8) /* 8 bits/char */
#define BMAP_SET(bmap, bit) ((bmap)[NODE_ARRAY_INDEX(bit)] |= 1 << NODE_ARRAY_OFFSET(bit))
#define BMAP_TEST(bmap, bit) ((bmap)[NODE_ARRAY_INDEX(bit)] & (1 << NODE_ARRAY_OFFSET(bit)))
/* bitmap length; _PXM is at most 255 */
#define PXM_BITMAP_LEN (MAX_PXM_DOMAINS / 8)
static u8 pxm_bitmap[PXM_BITMAP_LEN]; /* bitmap of proximity domains */
#define MAX_CHUNKS_PER_NODE 3
#define MAXCHUNKS (MAX_CHUNKS_PER_NODE * MAX_NUMNODES)
struct node_memory_chunk_s {
unsigned long start_pfn;
unsigned long end_pfn;
u8 pxm; // proximity domain of node
u8 nid; // which cnode contains this chunk?
u8 bank; // which mem bank on this node
};
static struct node_memory_chunk_s node_memory_chunk[MAXCHUNKS];
static int num_memory_chunks; /* total number of memory chunks */
static u8 __initdata apicid_to_pxm[MAX_APICID];
/* Identify CPU proximity domains */
static void __init parse_cpu_affinity_structure(char *p)
{
struct acpi_srat_cpu_affinity *cpu_affinity =
(struct acpi_srat_cpu_affinity *) p;
if ((cpu_affinity->flags & ACPI_SRAT_CPU_ENABLED) == 0)
return; /* empty entry */
/* mark this node as "seen" in node bitmap */
BMAP_SET(pxm_bitmap, cpu_affinity->proximity_domain_lo);
apicid_to_pxm[cpu_affinity->apic_id] = cpu_affinity->proximity_domain_lo;
printk("CPU 0x%02X in proximity domain 0x%02X\n",
cpu_affinity->apic_id, cpu_affinity->proximity_domain_lo);
}
/*
* Identify memory proximity domains and hot-remove capabilities.
* Fill node memory chunk list structure.
*/
static void __init parse_memory_affinity_structure (char *sratp)
{
unsigned long long paddr, size;
unsigned long start_pfn, end_pfn;
u8 pxm;
struct node_memory_chunk_s *p, *q, *pend;
struct acpi_srat_mem_affinity *memory_affinity =
(struct acpi_srat_mem_affinity *) sratp;
if ((memory_affinity->flags & ACPI_SRAT_MEM_ENABLED) == 0)
return; /* empty entry */
pxm = memory_affinity->proximity_domain & 0xff;
/* mark this node as "seen" in node bitmap */
BMAP_SET(pxm_bitmap, pxm);
/* calculate info for memory chunk structure */
paddr = memory_affinity->base_address;
size = memory_affinity->length;
start_pfn = paddr >> PAGE_SHIFT;
end_pfn = (paddr + size) >> PAGE_SHIFT;
if (num_memory_chunks >= MAXCHUNKS) {
printk("Too many mem chunks in SRAT. Ignoring %lld MBytes at %llx\n",
size/(1024*1024), paddr);
return;
}
/* Insertion sort based on base address */
pend = &node_memory_chunk[num_memory_chunks];
for (p = &node_memory_chunk[0]; p < pend; p++) {
if (start_pfn < p->start_pfn)
break;
}
if (p < pend) {
for (q = pend; q >= p; q--)
*(q + 1) = *q;
}
p->start_pfn = start_pfn;
p->end_pfn = end_pfn;
p->pxm = pxm;
num_memory_chunks++;
printk("Memory range 0x%lX to 0x%lX (type 0x%X) in proximity domain 0x%02X %s\n",
start_pfn, end_pfn,
memory_affinity->memory_type,
pxm,
((memory_affinity->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE) ?
"enabled and removable" : "enabled" ) );
}
/*
* The SRAT table always lists ascending addresses, so can always
* assume that the first "start" address that you see is the real
* start of the node, and that the current "end" address is after
* the previous one.
*/
static __init void node_read_chunk(int nid, struct node_memory_chunk_s *memory_chunk)
{
/*
* Only add present memory as told by the e820.
* There is no guarantee from the SRAT that the memory it
* enumerates is present at boot time because it represents
* *possible* memory hotplug areas the same as normal RAM.
*/
if (memory_chunk->start_pfn >= max_pfn) {
printk (KERN_INFO "Ignoring SRAT pfns: 0x%08lx -> %08lx\n",
memory_chunk->start_pfn, memory_chunk->end_pfn);
return;
}
if (memory_chunk->nid != nid)
return;
if (!node_has_online_mem(nid))
node_start_pfn[nid] = memory_chunk->start_pfn;
if (node_start_pfn[nid] > memory_chunk->start_pfn)
node_start_pfn[nid] = memory_chunk->start_pfn;
if (node_end_pfn[nid] < memory_chunk->end_pfn)
node_end_pfn[nid] = memory_chunk->end_pfn;
}
/* Parse the ACPI Static Resource Affinity Table */
static int __init acpi20_parse_srat(struct acpi_table_srat *sratp)
{
u8 *start, *end, *p;
int i, j, nid;
start = (u8 *)(&(sratp->reserved) + 1); /* skip header */
p = start;
end = (u8 *)sratp + sratp->header.length;
memset(pxm_bitmap, 0, sizeof(pxm_bitmap)); /* init proximity domain bitmap */
memset(node_memory_chunk, 0, sizeof(node_memory_chunk));
num_memory_chunks = 0;
while (p < end) {
switch (*p) {
case ACPI_SRAT_TYPE_CPU_AFFINITY:
parse_cpu_affinity_structure(p);
break;
case ACPI_SRAT_TYPE_MEMORY_AFFINITY:
parse_memory_affinity_structure(p);
break;
default:
printk("ACPI 2.0 SRAT: unknown entry skipped: type=0x%02X, len=%d\n", p[0], p[1]);
break;
}
p += p[1];
if (p[1] == 0) {
printk("acpi20_parse_srat: Entry length value is zero;"
" can't parse any further!\n");
break;
}
}
if (num_memory_chunks == 0) {
printk("could not finy any ACPI SRAT memory areas.\n");
goto out_fail;
}
/* Calculate total number of nodes in system from PXM bitmap and create
* a set of sequential node IDs starting at zero. (ACPI doesn't seem
* to specify the range of _PXM values.)
*/
/*
* MCD - we no longer HAVE to number nodes sequentially. PXM domain
* numbers could go as high as 256, and MAX_NUMNODES for i386 is typically
* 32, so we will continue numbering them in this manner until MAX_NUMNODES
* approaches MAX_PXM_DOMAINS for i386.
*/
nodes_clear(node_online_map);
for (i = 0; i < MAX_PXM_DOMAINS; i++) {
if (BMAP_TEST(pxm_bitmap, i)) {
int nid = acpi_map_pxm_to_node(i);
node_set_online(nid);
}
}
BUG_ON(num_online_nodes() == 0);
/* set cnode id in memory chunk structure */
for (i = 0; i < num_memory_chunks; i++)
node_memory_chunk[i].nid = pxm_to_node(node_memory_chunk[i].pxm);
printk("pxm bitmap: ");
for (i = 0; i < sizeof(pxm_bitmap); i++) {
printk("%02X ", pxm_bitmap[i]);
}
printk("\n");
printk("Number of logical nodes in system = %d\n", num_online_nodes());
printk("Number of memory chunks in system = %d\n", num_memory_chunks);
for (i = 0; i < MAX_APICID; i++)
apicid_2_node[i] = pxm_to_node(apicid_to_pxm[i]);
for (j = 0; j < num_memory_chunks; j++){
struct node_memory_chunk_s * chunk = &node_memory_chunk[j];
printk("chunk %d nid %d start_pfn %08lx end_pfn %08lx\n",
j, chunk->nid, chunk->start_pfn, chunk->end_pfn);
node_read_chunk(chunk->nid, chunk);
add_active_range(chunk->nid, chunk->start_pfn, chunk->end_pfn);
}
for_each_online_node(nid) {
unsigned long start = node_start_pfn[nid];
unsigned long end = node_end_pfn[nid];
memory_present(nid, start, end);
node_remap_size[nid] = node_memmap_size_bytes(nid, start, end);
}
return 1;
out_fail:
return 0;
}
struct acpi_static_rsdt {
struct acpi_table_rsdt table;
u32 padding[32]; /* Allow for 32 more table entries */
};
int __init get_memcfg_from_srat(void)
{
struct acpi_table_header *header = NULL;
struct acpi_table_rsdp *rsdp = NULL;
struct acpi_table_rsdt *rsdt = NULL;
acpi_native_uint rsdp_address = 0;
struct acpi_static_rsdt saved_rsdt;
int tables = 0;
int i = 0;
rsdp_address = acpi_os_get_root_pointer();
if (!rsdp_address) {
printk("%s: System description tables not found\n",
__func__);
goto out_err;
}
printk("%s: assigning address to rsdp\n", __func__);
rsdp = (struct acpi_table_rsdp *)(u32)rsdp_address;
if (!rsdp) {
printk("%s: Didn't find ACPI root!\n", __func__);
goto out_err;
}
printk(KERN_INFO "%.8s v%d [%.6s]\n", rsdp->signature, rsdp->revision,
rsdp->oem_id);
if (strncmp(rsdp->signature, ACPI_SIG_RSDP,strlen(ACPI_SIG_RSDP))) {
printk(KERN_WARNING "%s: RSDP table signature incorrect\n", __func__);
goto out_err;
}
rsdt = (struct acpi_table_rsdt *)
early_ioremap(rsdp->rsdt_physical_address, sizeof(saved_rsdt));
if (!rsdt) {
printk(KERN_WARNING
"%s: ACPI: Invalid root system description tables (RSDT)\n",
__func__);
goto out_err;
}
header = &rsdt->header;
if (strncmp(header->signature, ACPI_SIG_RSDT, strlen(ACPI_SIG_RSDT))) {
printk(KERN_WARNING "ACPI: RSDT signature incorrect\n");
early_iounmap(rsdt, sizeof(saved_rsdt));
goto out_err;
}
/*
* The number of tables is computed by taking the
* size of all entries (header size minus total
* size of RSDT) divided by the size of each entry
* (4-byte table pointers).
*/
tables = (header->length - sizeof(struct acpi_table_header)) / sizeof(u32);
if (!tables)
goto out_err;
memcpy(&saved_rsdt, rsdt, sizeof(saved_rsdt));
early_iounmap(rsdt, sizeof(saved_rsdt));
if (saved_rsdt.table.header.length > sizeof(saved_rsdt)) {
printk(KERN_WARNING "ACPI: Too big length in RSDT: %d\n",
saved_rsdt.table.header.length);
goto out_err;
}
printk("Begin SRAT table scan....%d\n", tables);
for (i = 0; i < tables; i++){
int result;
u32 length;
/* Map in header, then map in full table length. */
header = (struct acpi_table_header *)
early_ioremap(saved_rsdt.table.table_offset_entry[i], sizeof(struct acpi_table_header));
if (!header)
break;
printk(KERN_INFO "ACPI: %4.4s %08lX, %04X\n",
header->signature,
(unsigned long)saved_rsdt.table.table_offset_entry[i],
header->length);
if (strncmp((char *) &header->signature, ACPI_SIG_SRAT, 4)) {
early_iounmap(header, sizeof(struct acpi_table_header));
continue;
}
length = header->length;
early_iounmap(header, sizeof(struct acpi_table_header));
header = (struct acpi_table_header *)
early_ioremap(saved_rsdt.table.table_offset_entry[i], length);
if (!header)
break;
/* we've found the srat table. don't need to look at any more tables */
result = acpi20_parse_srat((struct acpi_table_srat *)header);
early_iounmap(header, length);
return result;
}
out_err:
remove_all_active_ranges();
printk("failed to get NUMA memory information from SRAT table\n");
return 0;
}