1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
|
// SPDX-License-Identifier: GPL-2.0
/*
* linux/arch/alpha/mm/numa.c
*
* DISCONTIGMEM NUMA alpha support.
*
* Copyright (C) 2001 Andrea Arcangeli <andrea@suse.de> SuSE
*/
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/memblock.h>
#include <linux/swap.h>
#include <linux/initrd.h>
#include <linux/pfn.h>
#include <linux/module.h>
#include <asm/hwrpb.h>
#include <asm/sections.h>
pg_data_t node_data[MAX_NUMNODES];
EXPORT_SYMBOL(node_data);
#undef DEBUG_DISCONTIG
#ifdef DEBUG_DISCONTIG
#define DBGDCONT(args...) printk(args)
#else
#define DBGDCONT(args...)
#endif
#define for_each_mem_cluster(memdesc, _cluster, i) \
for ((_cluster) = (memdesc)->cluster, (i) = 0; \
(i) < (memdesc)->numclusters; (i)++, (_cluster)++)
static void __init show_mem_layout(void)
{
struct memclust_struct * cluster;
struct memdesc_struct * memdesc;
int i;
/* Find free clusters, and init and free the bootmem accordingly. */
memdesc = (struct memdesc_struct *)
(hwrpb->mddt_offset + (unsigned long) hwrpb);
printk("Raw memory layout:\n");
for_each_mem_cluster(memdesc, cluster, i) {
printk(" memcluster %2d, usage %1lx, start %8lu, end %8lu\n",
i, cluster->usage, cluster->start_pfn,
cluster->start_pfn + cluster->numpages);
}
}
static void __init
setup_memory_node(int nid, void *kernel_end)
{
extern unsigned long mem_size_limit;
struct memclust_struct * cluster;
struct memdesc_struct * memdesc;
unsigned long start_kernel_pfn, end_kernel_pfn;
unsigned long start, end;
unsigned long node_pfn_start, node_pfn_end;
unsigned long node_min_pfn, node_max_pfn;
int i;
int show_init = 0;
/* Find the bounds of current node */
node_pfn_start = (node_mem_start(nid)) >> PAGE_SHIFT;
node_pfn_end = node_pfn_start + (node_mem_size(nid) >> PAGE_SHIFT);
/* Find free clusters, and init and free the bootmem accordingly. */
memdesc = (struct memdesc_struct *)
(hwrpb->mddt_offset + (unsigned long) hwrpb);
/* find the bounds of this node (node_min_pfn/node_max_pfn) */
node_min_pfn = ~0UL;
node_max_pfn = 0UL;
for_each_mem_cluster(memdesc, cluster, i) {
/* Bit 0 is console/PALcode reserved. Bit 1 is
non-volatile memory -- we might want to mark
this for later. */
if (cluster->usage & 3)
continue;
start = cluster->start_pfn;
end = start + cluster->numpages;
if (start >= node_pfn_end || end <= node_pfn_start)
continue;
if (!show_init) {
show_init = 1;
printk("Initializing bootmem allocator on Node ID %d\n", nid);
}
printk(" memcluster %2d, usage %1lx, start %8lu, end %8lu\n",
i, cluster->usage, cluster->start_pfn,
cluster->start_pfn + cluster->numpages);
if (start < node_pfn_start)
start = node_pfn_start;
if (end > node_pfn_end)
end = node_pfn_end;
if (start < node_min_pfn)
node_min_pfn = start;
if (end > node_max_pfn)
node_max_pfn = end;
}
if (mem_size_limit && node_max_pfn > mem_size_limit) {
static int msg_shown = 0;
if (!msg_shown) {
msg_shown = 1;
printk("setup: forcing memory size to %ldK (from %ldK).\n",
mem_size_limit << (PAGE_SHIFT - 10),
node_max_pfn << (PAGE_SHIFT - 10));
}
node_max_pfn = mem_size_limit;
}
if (node_min_pfn >= node_max_pfn)
return;
/* Update global {min,max}_low_pfn from node information. */
if (node_min_pfn < min_low_pfn)
min_low_pfn = node_min_pfn;
if (node_max_pfn > max_low_pfn)
max_pfn = max_low_pfn = node_max_pfn;
#if 0 /* we'll try this one again in a little while */
/* Cute trick to make sure our local node data is on local memory */
node_data[nid] = (pg_data_t *)(__va(node_min_pfn << PAGE_SHIFT));
#endif
printk(" Detected node memory: start %8lu, end %8lu\n",
node_min_pfn, node_max_pfn);
DBGDCONT(" DISCONTIG: node_data[%d] is at 0x%p\n", nid, NODE_DATA(nid));
/* Find the bounds of kernel memory. */
start_kernel_pfn = PFN_DOWN(KERNEL_START_PHYS);
end_kernel_pfn = PFN_UP(virt_to_phys(kernel_end));
if (!nid && (node_max_pfn < end_kernel_pfn || node_min_pfn > start_kernel_pfn))
panic("kernel loaded out of ram");
memblock_add_node(PFN_PHYS(node_min_pfn),
(node_max_pfn - node_min_pfn) << PAGE_SHIFT, nid);
/* Zone start phys-addr must be 2^(MAX_ORDER-1) aligned.
Note that we round this down, not up - node memory
has much larger alignment than 8Mb, so it's safe. */
node_min_pfn &= ~((1UL << (MAX_ORDER-1))-1);
NODE_DATA(nid)->node_start_pfn = node_min_pfn;
NODE_DATA(nid)->node_present_pages = node_max_pfn - node_min_pfn;
node_set_online(nid);
}
void __init
setup_memory(void *kernel_end)
{
unsigned long kernel_size;
int nid;
show_mem_layout();
nodes_clear(node_online_map);
min_low_pfn = ~0UL;
max_low_pfn = 0UL;
for (nid = 0; nid < MAX_NUMNODES; nid++)
setup_memory_node(nid, kernel_end);
kernel_size = virt_to_phys(kernel_end) - KERNEL_START_PHYS;
memblock_reserve(KERNEL_START_PHYS, kernel_size);
#ifdef CONFIG_BLK_DEV_INITRD
initrd_start = INITRD_START;
if (initrd_start) {
extern void *move_initrd(unsigned long);
initrd_end = initrd_start+INITRD_SIZE;
printk("Initial ramdisk at: 0x%p (%lu bytes)\n",
(void *) initrd_start, INITRD_SIZE);
if ((void *)initrd_end > phys_to_virt(PFN_PHYS(max_low_pfn))) {
if (!move_initrd(PFN_PHYS(max_low_pfn)))
printk("initrd extends beyond end of memory "
"(0x%08lx > 0x%p)\ndisabling initrd\n",
initrd_end,
phys_to_virt(PFN_PHYS(max_low_pfn)));
} else {
nid = kvaddr_to_nid(initrd_start);
memblock_reserve(virt_to_phys((void *)initrd_start),
INITRD_SIZE);
}
}
#endif /* CONFIG_BLK_DEV_INITRD */
}
void __init paging_init(void)
{
unsigned long max_zone_pfn[MAX_NR_ZONES] = {0, };
unsigned long dma_local_pfn;
/*
* The old global MAX_DMA_ADDRESS per-arch API doesn't fit
* in the NUMA model, for now we convert it to a pfn and
* we interpret this pfn as a local per-node information.
* This issue isn't very important since none of these machines
* have legacy ISA slots anyways.
*/
dma_local_pfn = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT;
max_zone_pfn[ZONE_DMA] = dma_local_pfn;
max_zone_pfn[ZONE_NORMAL] = max_pfn;
free_area_init(max_zone_pfn);
/* Initialize the kernel's ZERO_PGE. */
memset((void *)ZERO_PGE, 0, PAGE_SIZE);
}
|