[PATCH 2/3] x86 align highmem zone boundries with NUMA

From: Andy Whitcroft
Date: Tue May 09 2006 - 07:07:37 EST


x86 align highmem zone boundries with NUMA

When in x86 NUMA mode we allocate struct pages's node local and map
them into the kernel virtual address space in the remap space.
This space cuts into the end of ZONE_NORMAL. When we round
ZONE_NORMAL down we must ensure we maintain the zone boundry
constraint, we must round to MAX_ORDER.

Signed-off-by: Andy Whitcroft <apw@xxxxxxxxxxxx>
---
discontig.c | 5 ++++-
1 files changed, 4 insertions(+), 1 deletion(-)
diff -upN reference/arch/i386/mm/discontig.c current/arch/i386/mm/discontig.c
--- reference/arch/i386/mm/discontig.c
+++ current/arch/i386/mm/discontig.c
@@ -304,10 +304,13 @@ unsigned long __init setup_memory(void)
/* partially used pages are not usable - thus round upwards */
system_start_pfn = min_low_pfn = PFN_UP(init_pg_tables_end);

- system_max_low_pfn = max_low_pfn = find_max_low_pfn() - reserve_pages;
+ max_low_pfn = find_max_low_pfn() - reserve_pages;
printk("reserve_pages = %ld find_max_low_pfn() ~ %ld\n",
reserve_pages, max_low_pfn + reserve_pages);
printk("max_pfn = %ld\n", max_pfn);
+
+ system_max_low_pfn = max_low_pfn = zone_boundry_align_pfn(max_low_pfn);
+
#ifdef CONFIG_HIGHMEM
highstart_pfn = highend_pfn = max_pfn;
if (max_pfn > system_max_low_pfn)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/