Hi,
Currently, I'm testing memory hot add on a big machine which supports
acpi memory hotplug.
I found there is the case which Memory Device's _CRS contains plural
resources. This is because of requirements of Windows.
Current code cannot manage plural _CRS. This patch fixes it.
-- Kame
==
This patch fixes BUG in acpi memory hotplug.
Current acpi memory hotplug just look into the first entry of resources
in _CRS. But, _CRS can contain plural resources.
So, if _CRS contains plural resoureces, acpi memory hot add cannot add
all memory.
By this patch, acpi memory hotplug can deal with Memory Device, whose
_CRS contains plural resources.
Tested on ia64 memory hotplug test envrionment (not emulation, uses
alpha version firmware which supports dynamic reconfiguration of NUMA.)
Note: Microsoft's Windows Server 2003 requires big (>4G)resoureces to be
divided into small (<4G) resources. looks crazy, but not invalid.
(See http://www.microsoft.com/whdc/system/pnppwr/hotadd/hotaddmem.mspx)
For this reason, a firmware vendor who supports Windows writes plural
resoureces in a _CRS even if they are contiguous.
Signed-Off-By: Kenji Kaneshige <[EMAIL PROTECTED]>
Signed-Off-By: KAMEZAWA Hiroyuki <[EMAIL PROTECTED]>
---
drivers/acpi/acpi_memhotplug.c | 100 +++++++++++++++++++++++++++++------------
1 files changed, 73 insertions(+), 27 deletions(-)
Index: linux-2.6.16-rc5-mm2/drivers/acpi/acpi_memhotplug.c
===================================================================
Index: linux-2.6.16-rc5-mm3/drivers/acpi/acpi_memhotplug.c
===================================================================
--- linux-2.6.16-rc5-mm3.orig/drivers/acpi/acpi_memhotplug.c 2006-03-09
14:56:16.000000000 +0900
+++ linux-2.6.16-rc5-mm3/drivers/acpi/acpi_memhotplug.c 2006-03-09
14:58:29.000000000 +0900
@@ -68,44 +68,76 @@
},
};
+struct acpi_memory_info {
+ struct list_head list;
+ u64 start_addr; /* Memory Range start physical addr */
+ u64 length; /* Memory Range length */
+ unsigned short caching; /* memory cache attribute */
+ unsigned short write_protect; /* memory read/write attribute */
+ unsigned int enabled:1;
+};
+
struct acpi_memory_device {
acpi_handle handle;
unsigned int state; /* State of the memory device */
- unsigned short caching; /* memory cache attribute */
- unsigned short write_protect; /* memory read/write attribute */
- u64 start_addr; /* Memory Range start physical addr */
- u64 length; /* Memory Range length */
+ struct list_head res_list;
};
+static acpi_status
+acpi_memory_get_resource(struct acpi_resource *resource, void *context)
+{
+ struct acpi_memory_device *mem_device = context;
+ struct acpi_resource_address64 address64;
+ struct acpi_memory_info *info, *new;
+ acpi_status status;
+
+ status = acpi_resource_to_address64(resource, &address64);
+ if (ACPI_FAILURE(status) ||
+ (address64.resource_type != ACPI_MEMORY_RANGE))
+ return AE_OK;
+
+ list_for_each_entry(info, &mem_device->res_list, list) {
+ /* Can we combine the resource range information? */
+ if ((info->caching == address64.info.mem.caching) &&
+ (info->write_protect == address64.info.mem.write_protect) &&
+ (info->start_addr + info->length == address64.minimum)) {
+ info->length += address64.address_length;
+ return AE_OK;
+ }
+ }
+
+ new = acpi_os_allocate(sizeof(struct acpi_memory_info));
+ if (!new)
+ return AE_ERROR;
+
+ memset(new, 0, sizeof(*new));
+ INIT_LIST_HEAD(&new->list);
+ new->caching = address64.info.mem.caching;
+ new->write_protect = address64.info.mem.write_protect;
+ new->start_addr = address64.minimum;
+ new->length = address64.address_length;
+ list_add_tail(&new->list, &mem_device->res_list);
+
+ return AE_OK;
+}
+
static int
acpi_memory_get_device_resources(struct acpi_memory_device *mem_device)
{
acpi_status status;
- struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL };
- struct acpi_resource *resource = NULL;
- struct acpi_resource_address64 address64;
+ struct acpi_memory_info *info, *n;
ACPI_FUNCTION_TRACE("acpi_memory_get_device_resources");
/* Get the range from the _CRS */
- status = acpi_get_current_resources(mem_device->handle, &buffer);
- if (ACPI_FAILURE(status))
+ status = acpi_walk_resources(mem_device->handle, METHOD_NAME__CRS,
+ acpi_memory_get_resource, mem_device);
+ if (ACPI_FAILURE(status)) {
+ list_for_each_entry_safe(info, n, &mem_device->res_list, list)
+ acpi_os_free(info);
return_VALUE(-EINVAL);
-
- resource = (struct acpi_resource *)buffer.pointer;
- status = acpi_resource_to_address64(resource, &address64);
- if (ACPI_SUCCESS(status)) {
- if (address64.resource_type == ACPI_MEMORY_RANGE) {
- /* Populate the structure */
- mem_device->caching = address64.info.mem.caching;
- mem_device->write_protect =
- address64.info.mem.write_protect;
- mem_device->start_addr = address64.minimum;
- mem_device->length = address64.address_length;
- }
}
- acpi_os_free(buffer.pointer);
return_VALUE(0);
}
@@ -180,7 +212,8 @@
static int acpi_memory_enable_device(struct acpi_memory_device *mem_device)
{
- int result;
+ int result, num_enabled = 0;
+ struct acpi_memory_info *info;
ACPI_FUNCTION_TRACE("acpi_memory_enable_device");
@@ -195,12 +228,21 @@
/*
* Tell the VM there is more memory here...
* Note: Assume that this function returns zero on success
+ * We don't have memory-hot-add rollback function,now.
+ * (i.e. memory-hot-remove function)
*/
- result = add_memory(mem_device->start_addr, mem_device->length);
- if (result) {
+ list_for_each_entry(info, &mem_device->res_list, list) {
+ result = add_memory(info->start_addr, info->length);
+ if (result)
+ continue;
+ info->enabled = 1;
+ num_enabled++;
+ }
+
+ if (!num_enabled) {
ACPI_ERROR((AE_INFO, "add_memory failed"));
mem_device->state = MEMORY_INVALID_STATE;
- return result;
+ return -EINVAL;
}
return result;
@@ -244,8 +286,7 @@
static int acpi_memory_disable_device(struct acpi_memory_device *mem_device)
{
int result;
- u64 start = mem_device->start_addr;
- u64 len = mem_device->length;
+ struct acpi_memory_info *info, *n;
ACPI_FUNCTION_TRACE("acpi_memory_disable_device");
@@ -253,9 +294,14 @@
* Ask the VM to offline this memory range.
* Note: Assume that this function returns zero on success
*/
- result = remove_memory(start, len);
- if (result)
- return_VALUE(result);
+ list_for_each_entry_safe(info, n, &mem_device->res_list, list) {
+ if (info->enabled) {
+ result = remove_memory(info->start_addr, info->length);
+ if (result)
+ return_VALUE(result);
+ }
+ acpi_os_free(info);
+ }
/* Power-off and eject the device */
result = acpi_memory_powerdown_device(mem_device);
@@ -347,6 +393,7 @@
return_VALUE(-ENOMEM);
memset(mem_device, 0, sizeof(struct acpi_memory_device));
+ INIT_LIST_HEAD(&mem_device->res_list);
mem_device->handle = device->handle;
sprintf(acpi_device_name(device), "%s", ACPI_MEMORY_DEVICE_NAME);
sprintf(acpi_device_class(device), "%s", ACPI_MEMORY_DEVICE_CLASS);
-
To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
