The patch titled
wait_table and zonelist initializing for memory hotadd: wait_table initialization
has been removed from the -mm tree. Its filename is
wait_table-and-zonelist-initializing-for-memory-hotadd-wait_table-initialization.patch
This patch was dropped because it was merged into mainline or a subsystem tree
------------------------------------------------------
Subject: wait_table and zonelist initializing for memory hotadd: wait_table initialization
From: Yasunori Goto <y-goto@xxxxxxxxxxxxxx>
Wait_table is initialized according to zone size at boot time. But, we cannot
know the maixmum zone size when memory hotplug is enabled. It can be
changed.... And resizing of wait_table is hard.
So kernel allocate and initialzie wait_table as its maximum size.
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx>
Signed-off-by: Yasunori Goto <y-goto@xxxxxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxx>
---
mm/page_alloc.c | 59 +++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 53 insertions(+), 6 deletions(-)
diff -puN mm/page_alloc.c~wait_table-and-zonelist-initializing-for-memory-hotadd-wait_table-initialization mm/page_alloc.c
--- a/mm/page_alloc.c~wait_table-and-zonelist-initializing-for-memory-hotadd-wait_table-initialization
+++ a/mm/page_alloc.c
@@ -1727,6 +1727,7 @@ void __init build_all_zonelists(void)
*/
#define PAGES_PER_WAITQUEUE 256
+#ifndef CONFIG_MEMORY_HOTPLUG
static inline unsigned long wait_table_hash_nr_entries(unsigned long pages)
{
unsigned long size = 1;
@@ -1745,6 +1746,29 @@ static inline unsigned long wait_table_h
return max(size, 4UL);
}
+#else
+/*
+ * A zone's size might be changed by hot-add, so it is not possible to determine
+ * a suitable size for its wait_table. So we use the maximum size now.
+ *
+ * The max wait table size = 4096 x sizeof(wait_queue_head_t). ie:
+ *
+ * i386 (preemption config) : 4096 x 16 = 64Kbyte.
+ * ia64, x86-64 (no preemption): 4096 x 20 = 80Kbyte.
+ * ia64, x86-64 (preemption) : 4096 x 24 = 96Kbyte.
+ *
+ * The maximum entries are prepared when a zone's memory is (512K + 256) pages
+ * or more by the traditional way. (See above). It equals:
+ *
+ * i386, x86-64, powerpc(4K page size) : = ( 2G + 1M)byte.
+ * ia64(16K page size) : = ( 8G + 4M)byte.
+ * powerpc (64K page size) : = (32G +16M)byte.
+ */
+static inline unsigned long wait_table_hash_nr_entries(unsigned long pages)
+{
+ return 4096UL;
+}
+#endif
/*
* This is an integer logarithm so that shifts can be used later
@@ -2010,10 +2034,11 @@ void __init setup_per_cpu_pageset(void)
#endif
static __meminit
-void zone_wait_table_init(struct zone *zone, unsigned long zone_size_pages)
+int zone_wait_table_init(struct zone *zone, unsigned long zone_size_pages)
{
int i;
struct pglist_data *pgdat = zone->zone_pgdat;
+ size_t alloc_size;
/*
* The per-page waitqueue mechanism uses hashed waitqueues
@@ -2023,12 +2048,32 @@ void zone_wait_table_init(struct zone *z
wait_table_hash_nr_entries(zone_size_pages);
zone->wait_table_bits =
wait_table_bits(zone->wait_table_hash_nr_entries);
- zone->wait_table = (wait_queue_head_t *)
- alloc_bootmem_node(pgdat, zone->wait_table_hash_nr_entries
- * sizeof(wait_queue_head_t));
+ alloc_size = zone->wait_table_hash_nr_entries
+ * sizeof(wait_queue_head_t);
+
+ if (system_state == SYSTEM_BOOTING) {
+ zone->wait_table = (wait_queue_head_t *)
+ alloc_bootmem_node(pgdat, alloc_size);
+ } else {
+ /*
+ * This case means that a zone whose size was 0 gets new memory
+ * via memory hot-add.
+ * But it may be the case that a new node was hot-added. In
+ * this case vmalloc() will not be able to use this new node's
+ * memory - this wait_table must be initialized to use this new
+ * node itself as well.
+ * To use this new node's memory, further consideration will be
+ * necessary.
+ */
+ zone->wait_table = (wait_queue_head_t *)vmalloc(alloc_size);
+ }
+ if (!zone->wait_table)
+ return -ENOMEM;
for(i = 0; i < zone->wait_table_hash_nr_entries; ++i)
init_waitqueue_head(zone->wait_table + i);
+
+ return 0;
}
static __meminit void zone_pcp_init(struct zone *zone)
@@ -2055,8 +2100,10 @@ __meminit int init_currently_empty_zone(
unsigned long size)
{
struct pglist_data *pgdat = zone->zone_pgdat;
-
- zone_wait_table_init(zone, size);
+ int ret;
+ ret = zone_wait_table_init(zone, size);
+ if (ret)
+ return ret;
pgdat->nr_zones = zone_idx(zone) + 1;
zone->zone_start_pfn = zone_start_pfn;
_
Patches currently in -mm which might be from y-goto@xxxxxxxxxxxxxx are
origin.patch
catch-notification-of-memory-add-event-of-acpi-via-container-driver-register-start-func-for-memory-device.patch
catch-notification-of-memory-add-event-of-acpi-via-container-driveravoid-redundant-call-add_memory.patch
pgdat-allocation-for-new-node-add-specify-node-id.patch
pgdat-allocation-for-new-node-add-get-node-id-by-acpi.patch
pgdat-allocation-for-new-node-add-generic-alloc-node_data.patch
pgdat-allocation-for-new-node-add-refresh-node_data.patch
pgdat-allocation-for-new-node-add-export-kswapd-start-func.patch
pgdat-allocation-for-new-node-add-export-kswapd-start-func-fix.patch
pgdat-allocation-for-new-node-add-call-pgdat-allocation.patch
fix-compile-error-undefined-reference-for-sparc64.patch
register-sysfs-file-for-hotpluged-new-node.patch
pgdat-allocation-and-update-for-ia64-of-memory-hotplughold-pgdat-address-at-system-running.patch
pgdat-allocation-and-update-for-ia64-of-memory-hotplug-update-pgdat-address-array.patch
pgdat-allocation-and-update-for-ia64-of-memory-hotplugallocate-pgdat-and-per-node-data.patch
node-hotplug-register-cpu-remove-node-struct.patch
node-hotplug-register-cpu-remove-node-struct-alpha-fix.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
