This patch includes the add/remove/online functions for
memory hotplug as well as changes that allow for
init_memory_mapping() and friends to be used during
hot-add at runtime.
Include functions for adding and removing memory regions
in arch specified zones as well as a function for onlining
pages. This also makes the requisite changes (via liberal
use of the after_bootmem flag) to allow for reuse of the
init_memory_mapping() function to create and/or initialize
kernel pgtables for newly added regions.
Signed-off-by: Matt Tolentino <matthew.e.tolentino@intel.com>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
---
memhotplug-dave/arch/x86_64/mm/init.c | 123 +++++++++++++++++++++++++++++-----
1 files changed, 107 insertions(+), 16 deletions(-)
diff -puN arch/x86_64/mm/init.c~H6-add_remove_memory_functions_late_pgtable_init_code arch/x86_64/mm/init.c
--- memhotplug/arch/x86_64/mm/init.c~H6-add_remove_memory_functions_late_pgtable_init_code 2004-11-10 13:14:26.000000000 -0800
+++ memhotplug-dave/arch/x86_64/mm/init.c 2004-11-10 13:14:26.000000000 -0800
@@ -22,6 +22,8 @@
#include <linux/pagemap.h>
#include <linux/bootmem.h>
#include <linux/proc_fs.h>
+#include <linux/memory_hotplug.h>
+#include <linux/module.h>
#include <asm/processor.h>
#include <asm/system.h>
@@ -186,6 +188,12 @@ static __init void *alloc_low_page(int *
unsigned long pfn = table_end++, paddr;
void *adr;
+ if (after_bootmem) {
+ adr = (void *)get_zeroed_page(GFP_ATOMIC);
+ *phys = __pa(adr);
+ return adr;
+ }
+
if (pfn >= end_pfn)
panic("alloc_low_page: ran out of memory");
for (i = 0; temp_mappings[i].allocated; i++) {
@@ -205,12 +213,16 @@ static __init void *alloc_low_page(int *
static __init void unmap_low_page(int i)
{
- struct temp_map *ti = &temp_mappings[i];
+ struct temp_map *ti;
+
+ if (after_bootmem)
+ return;
+ ti = &temp_mappings[i];
set_pmd(ti->pmd, __pmd(0));
ti->allocated = 0;
}
-static void __init phys_pmd_init(pmd_t *pmd, unsigned long address, unsigned long end)
+static void phys_pmd_init(pmd_t *pmd, unsigned long address, unsigned long end)
{
int i;
@@ -228,11 +240,32 @@ static void __init phys_pmd_init(pmd_t *
}
}
-static void __init phys_pgd_init(pgd_t *pgd, unsigned long address, unsigned long end)
+static inline void
+phys_pmd_update(pgd_t *pgd, unsigned long address, unsigned long end)
+{
+ pmd_t *pmd = pmd_offset(pgd, (unsigned long)__va(address));
+
+ if (pmd_none(*pmd)) {
+ spin_lock(&init_mm.page_table_lock);
+ phys_pmd_init(pmd, address, end);
+ spin_unlock(&init_mm.page_table_lock);
+ __flush_tlb_all();
+ }
+}
+
+static void phys_pgd_init(pgd_t *pgd, unsigned long address, unsigned long end)
{
long i = pgd_index(address);
pgd = pgd + i;
+
+ if (after_bootmem) {
+ if (pgd_val(*pgd)) {
+ phys_pmd_update(pgd, address, end);
+ return;
+ }
+ }
+
for (; i < PTRS_PER_PGD; pgd++, i++) {
int map;
unsigned long paddr, pmd_phys;
@@ -240,19 +273,26 @@ static void __init phys_pgd_init(pgd_t *
paddr = (address & PML4_MASK) + i*PGDIR_SIZE;
if (paddr >= end) {
- for (; i < PTRS_PER_PGD; i++, pgd++)
+ for (; i < PTRS_PER_PGD; i++, pgd++) {
+ if (after_bootmem) spin_lock(&init_mm.page_table_lock);
set_pgd(pgd, __pgd(0));
+ if (after_bootmem) spin_unlock(&init_mm.page_table_lock);
+ }
break;
}
- if (!e820_mapped(paddr, paddr+PGDIR_SIZE, 0)) {
- set_pgd(pgd, __pgd(0));
- continue;
- }
+ if (!after_bootmem) {
+ if (!e820_mapped(paddr, paddr+PGDIR_SIZE, 0)) {
+ set_pgd(pgd, __pgd(0));
+ continue;
+ }
+ }
pmd = alloc_low_page(&map, &pmd_phys);
+ if (after_bootmem) spin_lock(&init_mm.page_table_lock);
set_pgd(pgd, __pgd(pmd_phys | _KERNPG_TABLE));
phys_pmd_init(pmd, paddr, end);
+ if (after_bootmem) spin_unlock(&init_mm.page_table_lock);
unmap_low_page(map);
}
__flush_tlb();
@@ -272,13 +312,17 @@ static void __init find_early_table_spac
table_start >>= PAGE_SHIFT;
table_end = table_start;
+
+ early_printk("kernel direct mapping tables upto %lx @ %lx-%lx\n", end,
+ table_start<<PAGE_SHIFT,
+ table_end<<PAGE_SHIFT);
}
/* Setup the direct mapping of the physical memory at PAGE_OFFSET.
This runs before bootmem is initialized and gets pages directly from the
physical memory. To access them they are temporarily mapped. */
-void __init init_memory_mapping(unsigned long start, unsigned long end)
+void init_memory_mapping(unsigned long start, unsigned long end)
{
unsigned long next;
@@ -290,7 +334,8 @@ void __init init_memory_mapping(unsigned
* mapped. Unfortunately this is done currently before the nodes are
* discovered.
*/
- find_early_table_space(end);
+ if (!after_bootmem)
+ find_early_table_space(end);
start += __PAGE_OFFSET;
end += __PAGE_OFFSET; /* turn virtual */
@@ -298,21 +343,24 @@ void __init init_memory_mapping(unsigned
for (; start < end; start = next) {
int map;
unsigned long pgd_phys;
- pgd_t *pgd = alloc_low_page(&map, &pgd_phys);
pml4_t *pml4 = pml4_offset_k(start);
+ pgd_t *pgd;
+ if (after_bootmem)
+ pgd = level3_offset_k(pml4, __PAGE_OFFSET);
+ else
+ pgd = alloc_low_page(&map, &pgd_phys);
next = start + PML4_SIZE;
if (next > end)
next = end;
phys_pgd_init(pgd, start-PAGE_OFFSET, end-PAGE_OFFSET);
- set_pml4(pml4, mk_kernel_pml4(pgd_phys));
+ if (!after_bootmem) /* matt - temporarily don't consider this */
+ set_pml4(pml4, mk_kernel_pml4(pgd_phys));
unmap_low_page(map);
}
- asm volatile("movq %%cr4,%0" : "=r" (mmu_cr4_features));
+ if (!after_bootmem)
+ asm volatile("movq %%cr4,%0" : "=r" (mmu_cr4_features));
__flush_tlb_all();
- early_printk("kernel direct mapping tables upto %lx @ %lx-%lx\n", end,
- table_start<<PAGE_SHIFT,
- table_end<<PAGE_SHIFT);
}
extern struct x8664_pda cpu_pda[NR_CPUS];
@@ -405,6 +453,49 @@ static inline int page_is_ram (unsigned
extern int swiotlb_force;
/*
+ * Memory hotplug specific functions
+ */
+
+#ifdef CONFIG_MEMORY_HOTPLUG
+
+void online_page(struct page *page)
+{
+ ClearPageReserved(page);
+ set_page_count(page, 1);
+ __free_page(page);
+ totalram_pages++;
+ num_physpages++;
+}
+
+int add_memory(u64 start, u64 size, unsigned long attr)
+{
+ struct pglist_data *pgdat = &contig_page_data;
+ struct zone *zone = pgdat->node_zones + MAX_NR_ZONES - 2;
+ unsigned long start_pfn = start >> PAGE_SHIFT;
+ unsigned long nr_pages = size >> PAGE_SHIFT;
+ int ret;
+
+ ret = __add_pages(zone, start_pfn, nr_pages, attr);
+ if (ret)
+ goto error;
+
+ init_memory_mapping(start, (start + size - 1));
+ return ret;
+error:
+ printk("%s: Problem encountered in __add_pages!\n", __func__);
+ return ret;
+}
+EXPORT_SYMBOL(add_memory);
+
+int remove_memory(u64 start, u64 size, unsigned long attr)
+{
+ return -ENOSYS;
+}
+EXPORT_SYMBOL(remove_memory);
+
+#endif
+
+/*
* devmem_is_allowed() checks to see if /dev/mem access to a certain address is
* valid. The argument is a physical page number.
*
_