| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <20210506152623.178731-4-zi.yan@sent.com>
Date: Thu, 6 May 2021 11:26:19 -0400
From: Zi Yan <zi.yan@...t.com>
To: David Hildenbrand <david@...hat.com>,
Oscar Salvador <osalvador@...e.de>
Cc: Michael Ellerman <mpe@...erman.id.au>,
Benjamin Herrenschmidt <benh@...nel.crashing.org>,
Thomas Gleixner <tglx@...utronix.de>, x86@...nel.org,
Andy Lutomirski <luto@...nel.org>,
"Rafael J . Wysocki" <rafael@...nel.org>,
Andrew Morton <akpm@...ux-foundation.org>,
Mike Rapoport <rppt@...nel.org>,
Anshuman Khandual <anshuman.khandual@....com>,
Michal Hocko <mhocko@...e.com>,
Dan Williams <dan.j.williams@...el.com>,
Wei Yang <richard.weiyang@...ux.alibaba.com>,
linux-ia64@...r.kernel.org, linux-kernel@...r.kernel.org,
linuxppc-dev@...ts.ozlabs.org, linux-mm@...ck.org,
Zi Yan <ziy@...dia.com>
Subject: [RFC PATCH 3/7] mm: memory_hotplug: decouple memory_block size with section size.
From: Zi Yan <ziy@...dia.com>
To enable subsection memory online/offline, we need to remove the
assumption of memory_block size being greater or equal to section size.
The following changes are made:
1. use (start_pfn, nr_pages) pair to specify memory_block size instead of
start_section_nr.
2. calculate memory_block id using phys / memory_block_size_bytes()
instead of section number.
The memory_block minimum size is set to the smaller of 128MB (the old
x86_64 section size) and section size instead.
Signed-off-by: Zi Yan <ziy@...dia.com>
---
drivers/base/memory.c | 176 ++++++++++++++++++++---------------------
drivers/base/node.c | 2 +-
include/linux/memory.h | 8 +-
mm/memory_hotplug.c | 6 +-
4 files changed, 98 insertions(+), 94 deletions(-)
diff --git a/drivers/base/memory.c b/drivers/base/memory.c
index b31b3af5c490..141431eb64a4 100644
--- a/drivers/base/memory.c
+++ b/drivers/base/memory.c
@@ -50,19 +50,15 @@ int mhp_online_type_from_str(const char *str)
static int sections_per_block;
-static inline unsigned long memory_block_id(unsigned long section_nr)
+static inline unsigned long phys_to_block_id(unsigned long phys)
{
- return section_nr / sections_per_block;
+ return phys / memory_block_size_bytes();
}
static inline unsigned long pfn_to_block_id(unsigned long pfn)
{
- return memory_block_id(pfn_to_section_nr(pfn));
-}
-
-static inline unsigned long phys_to_block_id(unsigned long phys)
-{
- return pfn_to_block_id(PFN_DOWN(phys));
+ /* calculate using memory_block_size_bytes() */
+ return phys_to_block_id(PFN_PHYS(pfn));
}
static int memory_subsys_online(struct device *dev);
@@ -118,7 +114,7 @@ static ssize_t phys_index_show(struct device *dev,
struct memory_block *mem = to_memory_block(dev);
unsigned long phys_index;
- phys_index = mem->start_section_nr / sections_per_block;
+ phys_index = pfn_to_section_nr(mem->start_pfn);
return sysfs_emit(buf, "%08lx\n", phys_index);
}
@@ -171,8 +167,8 @@ int memory_notify(unsigned long val, void *v)
static int memory_block_online(struct memory_block *mem)
{
- unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
- unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+ unsigned long start_pfn = mem->start_pfn;
+ unsigned long nr_pages = mem->nr_pages;
unsigned long nr_vmemmap_pages = mem->nr_vmemmap_pages;
struct zone *zone;
int ret;
@@ -212,8 +208,8 @@ static int memory_block_online(struct memory_block *mem)
static int memory_block_offline(struct memory_block *mem)
{
- unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
- unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+ unsigned long start_pfn = mem->start_pfn;
+ unsigned long nr_pages = mem->nr_pages;
unsigned long nr_vmemmap_pages = mem->nr_vmemmap_pages;
struct zone *zone;
int ret;
@@ -260,7 +256,7 @@ memory_block_action(struct memory_block *mem, unsigned long action)
break;
default:
WARN(1, KERN_WARNING "%s(%ld, %ld) unknown action: "
- "%ld\n", __func__, mem->start_section_nr, action, action);
+ "%ld\n", __func__, mem->start_pfn, mem->nr_pages, action);
ret = -EINVAL;
}
@@ -366,7 +362,7 @@ static ssize_t phys_device_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct memory_block *mem = to_memory_block(dev);
- unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
+ unsigned long start_pfn = mem->start_pfn;
return sysfs_emit(buf, "%d\n",
arch_get_memory_phys_device(start_pfn));
@@ -390,8 +386,8 @@ static ssize_t valid_zones_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct memory_block *mem = to_memory_block(dev);
- unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr);
- unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block;
+ unsigned long start_pfn = mem->start_pfn;
+ unsigned long nr_pages = mem->nr_pages;
struct zone *default_zone;
int len = 0;
int nid;
@@ -575,16 +571,6 @@ static struct memory_block *find_memory_block_by_id(unsigned long block_id)
return mem;
}
-/*
- * Called under device_hotplug_lock.
- */
-struct memory_block *find_memory_block(struct mem_section *section)
-{
- unsigned long block_id = memory_block_id(__section_nr(section));
-
- return find_memory_block_by_id(block_id);
-}
-
static struct attribute *memory_memblk_attrs[] = {
&dev_attr_phys_index.attr,
&dev_attr_state.attr,
@@ -614,7 +600,7 @@ int register_memory(struct memory_block *memory)
int ret;
memory->dev.bus = &memory_subsys;
- memory->dev.id = memory->start_section_nr / sections_per_block;
+ memory->dev.id = memory->start_pfn / (memory_block_size_bytes() >> PAGE_SHIFT);
memory->dev.release = memory_block_release;
memory->dev.groups = memory_memblk_attr_groups;
memory->dev.offline = memory->state == MEM_OFFLINE;
@@ -633,57 +619,89 @@ int register_memory(struct memory_block *memory)
return ret;
}
-static int init_memory_block(unsigned long block_id, unsigned long state,
+static void unregister_memory(struct memory_block *memory)
+{
+ if (WARN_ON_ONCE(memory->dev.bus != &memory_subsys))
+ return;
+
+ WARN_ON(xa_erase(&memory_blocks, memory->dev.id) == NULL);
+
+ /* drop the ref. we got via find_memory_block() */
+ put_device(&memory->dev);
+ device_unregister(&memory->dev);
+}
+
+static int init_memory_blocks(unsigned long start_pfn, unsigned long num_pages, unsigned long state,
unsigned long nr_vmemmap_pages)
{
struct memory_block *mem;
int ret = 0;
+ unsigned long block_nr_pages = memory_block_size_bytes() / PAGE_SIZE;
+ unsigned long block_start_pfn;
- mem = find_memory_block_by_id(block_id);
- if (mem) {
- put_device(&mem->dev);
- return -EEXIST;
- }
- mem = kzalloc(sizeof(*mem), GFP_KERNEL);
- if (!mem)
- return -ENOMEM;
-
- mem->start_section_nr = block_id * sections_per_block;
- mem->state = state;
- mem->nid = NUMA_NO_NODE;
- mem->nr_vmemmap_pages = nr_vmemmap_pages;
+ for (block_start_pfn = start_pfn; num_pages != 0; block_start_pfn += block_nr_pages) {
+ unsigned long block_id = pfn_to_block_id(block_start_pfn);
- ret = register_memory(mem);
-
- return ret;
+ mem = find_memory_block_by_id(block_id);
+ if (mem) {
+ put_device(&mem->dev);
+ return -EEXIST;
+ }
+ mem = kzalloc(sizeof(*mem), GFP_KERNEL);
+ if (!mem)
+ return -ENOMEM;
+
+ mem->start_pfn = block_start_pfn;
+ mem->nr_pages = min(num_pages, block_nr_pages);
+ mem->state = state;
+ mem->nid = NUMA_NO_NODE;
+ mem->nr_vmemmap_pages = nr_vmemmap_pages;
+
+ ret = register_memory(mem);
+
+ if (ret) {
+ unsigned long unregister_block_pfn;
+
+ for (unregister_block_pfn = start_pfn;
+ unregister_block_pfn < block_start_pfn;
+ unregister_block_pfn -= block_nr_pages) {
+ block_id = pfn_to_block_id(unregister_block_pfn);
+ mem = find_memory_block_by_id(block_id);
+ if (WARN_ON_ONCE(!mem))
+ continue;
+ unregister_memory(mem);
+ }
+ return -EINVAL;
+ }
+ if (num_pages > block_nr_pages)
+ num_pages -= block_nr_pages;
+ else
+ num_pages = 0;
+ }
+ return 0;
}
-static int add_memory_block(unsigned long base_section_nr)
+static void add_whole_section_memory_block(unsigned long base_section_nr)
{
- int section_count = 0;
- unsigned long nr;
+ int ret;
+ unsigned long start_pfn = section_nr_to_pfn(base_section_nr);
+ unsigned long nr_pages = 0;
+ struct mem_section *ms = __nr_to_section(base_section_nr);
- for (nr = base_section_nr; nr < base_section_nr + sections_per_block;
- nr++)
- if (present_section_nr(nr))
- section_count++;
+ if (bitmap_full(ms->usage->subsection_map, SUBSECTIONS_PER_SECTION))
+ nr_pages = PAGES_PER_SECTION;
+ else
+ nr_pages = PAGES_PER_SUBSECTION *
+ bitmap_weight(ms->usage->subsection_map, SUBSECTIONS_PER_SECTION);
- if (section_count == 0)
- return 0;
- return init_memory_block(memory_block_id(base_section_nr),
- MEM_ONLINE, 0);
-}
-static void unregister_memory(struct memory_block *memory)
-{
- if (WARN_ON_ONCE(memory->dev.bus != &memory_subsys))
+ if (!nr_pages)
return;
- WARN_ON(xa_erase(&memory_blocks, memory->dev.id) == NULL);
-
- /* drop the ref. we got via find_memory_block() */
- put_device(&memory->dev);
- device_unregister(&memory->dev);
+ ret = init_memory_blocks(start_pfn, nr_pages, MEM_ONLINE, 0);
+ if (ret)
+ panic("%s() failed to add memory block: %d\n", __func__,
+ ret);
}
/*
@@ -696,31 +714,16 @@ static void unregister_memory(struct memory_block *memory)
int create_memory_block_devices(unsigned long start, unsigned long size,
unsigned long vmemmap_pages)
{
- const unsigned long start_block_id = pfn_to_block_id(PFN_DOWN(start));
- unsigned long end_block_id = pfn_to_block_id(PFN_DOWN(start + size));
- struct memory_block *mem;
- unsigned long block_id;
+ unsigned long start_pfn = PFN_DOWN(start);
+ unsigned long end_pfn = PFN_DOWN(start + size);
int ret = 0;
if (WARN_ON_ONCE(!IS_ALIGNED(start, memory_block_size_bytes()) ||
!IS_ALIGNED(size, memory_block_size_bytes())))
return -EINVAL;
- for (block_id = start_block_id; block_id != end_block_id; block_id++) {
- ret = init_memory_block(block_id, MEM_OFFLINE, vmemmap_pages);
- if (ret)
- break;
- }
- if (ret) {
- end_block_id = block_id;
- for (block_id = start_block_id; block_id != end_block_id;
- block_id++) {
- mem = find_memory_block_by_id(block_id);
- if (WARN_ON_ONCE(!mem))
- continue;
- unregister_memory(mem);
- }
- }
+ ret = init_memory_blocks(start_pfn, end_pfn - start_pfn, MEM_OFFLINE, vmemmap_pages);
+
return ret;
}
@@ -807,10 +810,7 @@ void __init memory_dev_init(void)
*/
for (nr = 0; nr <= __highest_present_section_nr;
nr += sections_per_block) {
- ret = add_memory_block(nr);
- if (ret)
- panic("%s() failed to add memory block: %d\n", __func__,
- ret);
+ add_whole_section_memory_block(nr);
}
}
diff --git a/drivers/base/node.c b/drivers/base/node.c
index 2c36f61d30bc..76d67b8ddf1b 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -809,7 +809,7 @@ static int register_mem_block_under_node_early(struct memory_block *mem_blk,
void *arg)
{
unsigned long memory_block_pfns = memory_block_size_bytes() / PAGE_SIZE;
- unsigned long start_pfn = section_nr_to_pfn(mem_blk->start_section_nr);
+ unsigned long start_pfn = mem_blk->start_pfn;
unsigned long end_pfn = start_pfn + memory_block_pfns - 1;
int nid = *(int *)arg;
unsigned long pfn;
diff --git a/include/linux/memory.h b/include/linux/memory.h
index 97e92e8b556a..e9590c7c6a9e 100644
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -21,10 +21,15 @@
#include <linux/mutex.h>
#include <linux/notifier.h>
+#if SECTION_SIZE_BITS > 27 /* 128MB */
+#define MIN_MEMORY_BLOCK_SIZE (1UL << 27)
+#else
#define MIN_MEMORY_BLOCK_SIZE (1UL << SECTION_SIZE_BITS)
+#endif
struct memory_block {
- unsigned long start_section_nr;
+ unsigned long start_pfn;
+ unsigned long nr_pages;
unsigned long state; /* serialized by the dev->lock */
int online_type; /* for passing data to online routine */
int nid; /* NID for this memory block */
@@ -90,7 +95,6 @@ int create_memory_block_devices(unsigned long start, unsigned long size,
void remove_memory_block_devices(unsigned long start, unsigned long size);
extern void memory_dev_init(void);
extern int memory_notify(unsigned long val, void *v);
-extern struct memory_block *find_memory_block(struct mem_section *);
typedef int (*walk_memory_blocks_func_t)(struct memory_block *, void *);
extern int walk_memory_blocks(unsigned long start, unsigned long size,
void *arg, walk_memory_blocks_func_t func);
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 70620d0dd923..6e93b0ecc5cb 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1872,8 +1872,8 @@ static int check_memblock_offlined_cb(struct memory_block *mem, void *arg)
if (unlikely(ret)) {
phys_addr_t beginpa, endpa;
- beginpa = PFN_PHYS(section_nr_to_pfn(mem->start_section_nr));
- endpa = beginpa + memory_block_size_bytes() - 1;
+ beginpa = PFN_PHYS(mem->start_pfn);
+ endpa = beginpa + mem->nr_pages * PAGE_SIZE - 1;
pr_warn("removing memory fails, because memory [%pa-%pa] is onlined\n",
&beginpa, &endpa);
@@ -2079,7 +2079,7 @@ static int try_offline_memory_block(struct memory_block *mem, void *arg)
* with multiple zones within one memory block will be rejected
* by offlining code ... so we don't care about that.
*/
- page = pfn_to_online_page(section_nr_to_pfn(mem->start_section_nr));
+ page = pfn_to_online_page(mem->start_pfn);
if (page && zone_idx(page_zone(page)) == ZONE_MOVABLE)
online_type = MMOP_ONLINE_MOVABLE;
--
2.30.2
Powered by blists - more mailing lists