Sometimes, firmware may expose interleaved memory layout like this:
Early memory node ranges
node 1: [mem 0x0000000000000000-0x000000011fffffff]
node 2: [mem 0x0000000120000000-0x000000014fffffff]
node 1: [mem 0x0000000150000000-0x00000001ffffffff]
node 0: [mem 0x0000000200000000-0x000000048fffffff]
node 2: [mem 0x0000000490000000-0x00000007ffffffff]
In that case, we can see memory blocks assigned to multiple nodes in sysfs:
$ ls -l /sys/devices/system/memory/memory21
total 0
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
-rw-r--r-- 1 root root 65536 Aug 24 05:27 online
-r--r--r-- 1 root root 65536 Aug 24 05:27 phys_device
-r--r--r-- 1 root root 65536 Aug 24 05:27 phys_index
drwxr-xr-x 2 root root 0 Aug 24 05:27 power
-r--r--r-- 1 root root 65536 Aug 24 05:27 removable
-rw-r--r-- 1 root root 65536 Aug 24 05:27 state
lrwxrwxrwx 1 root root 0 Aug 24 05:25 subsystem -> ../../../../bus/memory
-rw-r--r-- 1 root root 65536 Aug 24 05:25 uevent
-r--r--r-- 1 root root 65536 Aug 24 05:27 valid_zones
The same applies in the node's directory with a memory21 link in both the
node1 and node2's directory.
This is wrong but doesn't prevent the system to run. However when later one
of these memory blocks is hot-unplugged and then hot-plugged, the system is
detecting an inconsistency in the sysfs layout and a BUG_ON() is raised:
------------[ cut here ]------------
kernel BUG at /Users/laurent/src/linux-ppc/mm/memory_hotplug.c:1084!
Oops: Exception in kernel mode, sig: 5 [#1]
LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
Modules linked in: rpadlpar_io rpaphp pseries_rng rng_core vmx_crypto gf128mul binfmt_misc ip_tables x_tables xfs libcrc32c crc32c_vpmsum autofs4
CPU: 8 PID: 10256 Comm: drmgr Not tainted 5.9.0-rc1+ #25
NIP: c000000000403f34 LR: c000000000403f2c CTR: 0000000000000000
REGS: c0000004876e3660 TRAP: 0700 Not tainted (5.9.0-rc1+)
MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 24000448 XER: 20040000
CFAR: c000000000846d20 IRQMASK: 0
GPR00: c000000000403f2c c0000004876e38f0 c0000000012f6f00 ffffffffffffffef
GPR04: 0000000000000227 c0000004805ae680 0000000000000000 00000004886f0000
GPR08: 0000000000000226 0000000000000003 0000000000000002 fffffffffffffffd
GPR12: 0000000088000484 c00000001ec96280 0000000000000000 0000000000000000
GPR16: 0000000000000000 0000000000000000 0000000000000004 0000000000000003
GPR20: c00000047814ffe0 c0000007ffff7c08 0000000000000010 c0000000013332c8
GPR24: 0000000000000000 c0000000011f6cc0 0000000000000000 0000000000000000
GPR28: ffffffffffffffef 0000000000000001 0000000150000000 0000000010000000
NIP [c000000000403f34] add_memory_resource+0x244/0x340
LR [c000000000403f2c] add_memory_resource+0x23c/0x340
Call Trace:
[c0000004876e38f0] [c000000000403f2c] add_memory_resource+0x23c/0x340 (unreliable)
[c0000004876e39c0] [c00000000040408c] __add_memory+0x5c/0xf0
[c0000004876e39f0] [c0000000000e2b94] dlpar_add_lmb+0x1b4/0x500
[c0000004876e3ad0] [c0000000000e3888] dlpar_memory+0x1f8/0xb80
[c0000004876e3b60] [c0000000000dc0d0] handle_dlpar_errorlog+0xc0/0x190
[c0000004876e3bd0] [c0000000000dc398] dlpar_store+0x198/0x4a0
[c0000004876e3c90] [c00000000072e630] kobj_attr_store+0x30/0x50
[c0000004876e3cb0] [c00000000051f954] sysfs_kf_write+0x64/0x90
[c0000004876e3cd0] [c00000000051ee40] kernfs_fop_write+0x1b0/0x290
[c0000004876e3d20] [c000000000438dd8] vfs_write+0xe8/0x290
[c0000004876e3d70] [c0000000004391ac] ksys_write+0xdc/0x130
[c0000004876e3dc0] [c000000000034e40] system_call_exception+0x160/0x270
[c0000004876e3e20] [c00000000000d740] system_call_common+0xf0/0x27c
Instruction dump:
48442e35 60000000 0b030000 3cbe0001 7fa3eb78 7bc48402 38a5fffe 7ca5fa14
78a58402 48442db1 60000000 7c7c1b78 <0b030000> 7f23cb78 4bda371d 60000000
---[ end trace 562fd6c109cd0fb2 ]---
This has been seen on PowerPC LPAR.
The root cause of this issue is that when node's memory is registered, the
range used can overlap another node's range, thus the memory block is
registered to multiple node in sysfs.
There are 2 issues here:
a. The sysfs memory and node's layout are broken due to these multiple
links
b. The link errors in link_mem_sections() should not lead to a system
panic.
To address a. register_mem_sect_under_node should not rely on the system
state to detect whether the link operation is triggered by a hot plug
operation or not. This is addressed by the patch 1 and 2 of this series.
The patch 3 is addressing the point b.
Laurent Dufour (3):
mm: replace memmap_context by memplug_context
mm: don't rely on system state to detect hot-plug operations
mm: don't panic when links can't be created in sysfs
arch/ia64/mm/init.c | 6 ++---
drivers/base/node.c | 40 +++++++++++++++++++++++++---------
include/linux/memory_hotplug.h | 9 ++++++++
include/linux/mm.h | 2 +-
include/linux/mmzone.h | 4 ----
include/linux/node.h | 14 +++++++-----
mm/memory_hotplug.c | 6 ++---
mm/page_alloc.c | 10 ++++-----
8 files changed, 59 insertions(+), 32 deletions(-)
--
2.28.0
The memmap_context is used to detect whether a memory operation is due to a
hot-add operation or happening at boot time.
Makes it general to the hotplug operation, renaming it at memplug_context
and move its define in the corresponding header.
There is no functional change introduced by this patch
Suggested-by: David Hildenbrand <[email protected]>
Signed-off-by: Laurent Dufour <[email protected]>
---
arch/ia64/mm/init.c | 6 +++---
include/linux/memory_hotplug.h | 9 +++++++++
include/linux/mm.h | 2 +-
include/linux/mmzone.h | 4 ----
mm/memory_hotplug.c | 2 +-
mm/page_alloc.c | 10 +++++-----
6 files changed, 19 insertions(+), 14 deletions(-)
diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c
index 0b3fb4c7af29..b5054b5e77c8 100644
--- a/arch/ia64/mm/init.c
+++ b/arch/ia64/mm/init.c
@@ -538,7 +538,7 @@ virtual_memmap_init(u64 start, u64 end, void *arg)
if (map_start < map_end)
memmap_init_zone((unsigned long)(map_end - map_start),
args->nid, args->zone, page_to_pfn(map_start),
- MEMMAP_EARLY, NULL);
+ MEMPLUG_EARLY, NULL);
return 0;
}
@@ -547,8 +547,8 @@ memmap_init (unsigned long size, int nid, unsigned long zone,
unsigned long start_pfn)
{
if (!vmem_map) {
- memmap_init_zone(size, nid, zone, start_pfn, MEMMAP_EARLY,
- NULL);
+ memmap_init_zone(size, nid, zone, start_pfn,
+ MEMPLUG_EARLY, NULL);
} else {
struct page *start;
struct memmap_init_callback_data args;
diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
index 375515803cd8..cd2bd21d3a4d 100644
--- a/include/linux/memory_hotplug.h
+++ b/include/linux/memory_hotplug.h
@@ -15,6 +15,15 @@ struct memory_block;
struct resource;
struct vmem_altmap;
+/*
+ * Memory plugin context, use to differentiate memory added at boot time and
+ * hot-plugged memory.
+ */
+enum memplug_context {
+ MEMPLUG_EARLY,
+ MEMPLUG_HOTPLUG,
+};
+
#ifdef CONFIG_MEMORY_HOTPLUG
/*
* Return page for the valid pfn only if the page is online. All pfn
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 1983e08f5906..4157cc1bd17f 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -2409,7 +2409,7 @@ extern int __meminit __early_pfn_to_nid(unsigned long pfn,
extern void set_dma_reserve(unsigned long new_dma_reserve);
extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long,
- enum memmap_context, struct vmem_altmap *);
+ enum memplug_context, struct vmem_altmap *);
extern void setup_per_zone_wmarks(void);
extern int __meminit init_per_zone_wmark_min(void);
extern void mem_init(void);
diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 8379432f4f2f..ec254ab793b5 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -824,10 +824,6 @@ bool zone_watermark_ok(struct zone *z, unsigned int order,
unsigned int alloc_flags);
bool zone_watermark_ok_safe(struct zone *z, unsigned int order,
unsigned long mark, int highest_zoneidx);
-enum memmap_context {
- MEMMAP_EARLY,
- MEMMAP_HOTPLUG,
-};
extern void init_currently_empty_zone(struct zone *zone, unsigned long start_pfn,
unsigned long size);
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index e9d5ab5d3ca0..fc21625e42de 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -729,7 +729,7 @@ void __ref move_pfn_range_to_zone(struct zone *zone, unsigned long start_pfn,
* are reserved so nobody should be touching them so we should be safe
*/
memmap_init_zone(nr_pages, nid, zone_idx(zone), start_pfn,
- MEMMAP_HOTPLUG, altmap);
+ MEMPLUG_HOTPLUG, altmap);
set_zone_contiguous(zone);
}
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index fab5e97dc9ca..7f49fcf38b8c 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -5975,7 +5975,7 @@ overlap_memmap_init(unsigned long zone, unsigned long *pfn)
* done. Non-atomic initialization, single-pass.
*/
void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
- unsigned long start_pfn, enum memmap_context context,
+ unsigned long start_pfn, enum memplug_context context,
struct vmem_altmap *altmap)
{
unsigned long pfn, end_pfn = start_pfn + size;
@@ -6007,7 +6007,7 @@ void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
* There can be holes in boot-time mem_map[]s handed to this
* function. They do not exist on hotplugged memory.
*/
- if (context == MEMMAP_EARLY) {
+ if (context == MEMPLUG_EARLY) {
if (overlap_memmap_init(zone, &pfn))
continue;
if (defer_init(nid, pfn, end_pfn))
@@ -6016,7 +6016,7 @@ void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
page = pfn_to_page(pfn);
__init_single_page(page, pfn, zone, nid);
- if (context == MEMMAP_HOTPLUG)
+ if (context == MEMPLUG_HOTPLUG)
__SetPageReserved(page);
/*
@@ -6099,7 +6099,7 @@ void __ref memmap_init_zone_device(struct zone *zone,
* check here not to call set_pageblock_migratetype() against
* pfn out of zone.
*
- * Please note that MEMMAP_HOTPLUG path doesn't clear memmap
+ * Please note that MEMPLUG_HOTPLUG path doesn't clear memmap
* because this is done early in section_activate()
*/
if (!(pfn & (pageblock_nr_pages - 1))) {
@@ -6137,7 +6137,7 @@ void __meminit __weak memmap_init(unsigned long size, int nid,
if (end_pfn > start_pfn) {
size = end_pfn - start_pfn;
memmap_init_zone(size, nid, zone, start_pfn,
- MEMMAP_EARLY, NULL);
+ MEMPLUG_EARLY, NULL);
}
}
}
--
2.28.0
Le 11/09/2020 à 16:59, David Hildenbrand a écrit :
>> struct page *start;
>> struct memmap_init_callback_data args;
>> diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
>> index 375515803cd8..cd2bd21d3a4d 100644
>> --- a/include/linux/memory_hotplug.h
>> +++ b/include/linux/memory_hotplug.h
>> @@ -15,6 +15,15 @@ struct memory_block;
>> struct resource;
>> struct vmem_altmap;
>>
>> +/*
>> + * Memory plugin context, use to differentiate memory added at boot time and
>
> "Memory plugin context" sounds weird.
Any suggestion ?
>> + * hot-plugged memory.
>> + */
>> +enum memplug_context {
>> + MEMPLUG_EARLY,
>> + MEMPLUG_HOTPLUG,
>> +};
>
> Not fully sure this is the right place, though. memory_hotplug.h is
> about MEMPLUG_HOTPLUG only ... I'd leave it at the old spot.
Fair enough.
At boot time, or when doing memory hot-add operations, if the links in
sysfs can't be created, the system is still able to run, so just report the
error in the kernel log.
Since the number of memory blocks managed could be high, the messages are
rate limited.
As a consequence, link_mem_sections() has no status to report anymore.
Signed-off-by: Laurent Dufour <[email protected]>
Cc: David Hildenbrand <[email protected]>
---
drivers/base/node.c | 25 +++++++++++++++++--------
include/linux/node.h | 17 ++++++++---------
mm/memory_hotplug.c | 5 ++---
3 files changed, 27 insertions(+), 20 deletions(-)
diff --git a/drivers/base/node.c b/drivers/base/node.c
index 862516c5a5ae..749a1c8ea992 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -811,12 +811,21 @@ static int register_mem_sect_under_node(struct memory_block *mem_blk,
ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
&mem_blk->dev.kobj,
kobject_name(&mem_blk->dev.kobj));
- if (ret)
- return ret;
+ if (ret && ret != -EEXIST)
+ pr_err_ratelimited(
+ "can't create %s to %s link in sysfs (%d)\n",
+ kobject_name(&node_devices[nid]->dev.kobj),
+ kobject_name(&mem_blk->dev.kobj), ret);
- return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
+ ret = sysfs_create_link_nowarn(&mem_blk->dev.kobj,
&node_devices[nid]->dev.kobj,
kobject_name(&node_devices[nid]->dev.kobj));
+ if (ret && ret != -EEXIST)
+ pr_err_ratelimited(
+ "can't create %s to %s link in sysfs (%d)\n",
+ kobject_name(&mem_blk->dev.kobj),
+ kobject_name(&node_devices[nid]->dev.kobj),
+ ret);
}
/* mem section does not span the specified node */
return 0;
@@ -837,17 +846,17 @@ void unregister_memory_block_under_nodes(struct memory_block *mem_blk)
kobject_name(&node_devices[mem_blk->nid]->dev.kobj));
}
-int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
- enum memplug_context context)
+void link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
+ enum memplug_context context)
{
struct rmsun_args args = {
.nid = nid,
.context = context,
};
- return walk_memory_blocks(PFN_PHYS(start_pfn),
- PFN_PHYS(end_pfn - start_pfn), (void *)&args,
- register_mem_sect_under_node);
+ walk_memory_blocks(PFN_PHYS(start_pfn),
+ PFN_PHYS(end_pfn - start_pfn), (void *)&args,
+ register_mem_sect_under_node);
}
#ifdef CONFIG_HUGETLBFS
diff --git a/include/linux/node.h b/include/linux/node.h
index 8ff08520488c..6bdd6f3ed3aa 100644
--- a/include/linux/node.h
+++ b/include/linux/node.h
@@ -99,15 +99,14 @@ extern struct node *node_devices[];
typedef void (*node_registration_func_t)(struct node *);
#if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) && defined(CONFIG_NUMA)
-extern int link_mem_sections(int nid, unsigned long start_pfn,
- unsigned long end_pfn,
- enum memplug_context context);
+void link_mem_sections(int nid, unsigned long start_pfn,
+ unsigned long end_pfn,
+ enum memplug_context context);
#else
-static inline int link_mem_sections(int nid, unsigned long start_pfn,
- unsigned long end_pfn,
- enum memplug_context context)
+static inline void link_mem_sections(int nid, unsigned long start_pfn,
+ unsigned long end_pfn,
+ enum memplug_context context)
{
- return 0;
}
#endif
@@ -130,8 +129,8 @@ static inline int register_one_node(int nid)
if (error)
return error;
/* link memory sections under this node */
- error = link_mem_sections(nid, start_pfn, end_pfn,
- MEMPLUG_EARLY);
+ link_mem_sections(nid, start_pfn, end_pfn,
+ MEMPLUG_EARLY);
}
return error;
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 912d355ca446..668418071a49 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1080,9 +1080,8 @@ int __ref add_memory_resource(int nid, struct resource *res)
}
/* link memory sections under this node.*/
- ret = link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1),
- MEMPLUG_HOTPLUG);
- BUG_ON(ret);
+ link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1),
+ MEMPLUG_HOTPLUG);
/* create new memmap entry */
if (!strcmp(res->name, "System RAM"))
--
2.28.0
In register_mem_sect_under_node() the system_state’s value is checked to
detect whether the operation the call is made during boot time or during an
hot-plug operation. Unfortunately, that check against SYSTEM_BOOTING is
wrong because regular memory is registered at SYSTEM_SCHEDULING state. In
addition memory hot-plug operation can be triggered at this system state
too by the ACPI. So checking against the system state is not enough.
The consequence is that on system with interleaved node's ranges like this:
Early memory node ranges
node 1: [mem 0x0000000000000000-0x000000011fffffff]
node 2: [mem 0x0000000120000000-0x000000014fffffff]
node 1: [mem 0x0000000150000000-0x00000001ffffffff]
node 0: [mem 0x0000000200000000-0x000000048fffffff]
node 2: [mem 0x0000000490000000-0x00000007ffffffff]
This can be seen on PowerPC LPAR after multiple memory hot-plug and
hot-unplug operations are done. At the next reboot the node's memory ranges
can be interleaved and since the call to link_mem_sections() is made in
topology_init() while the system is in the SYSTEM_SCHEDULING state, the
node's id is not checked, and the sections registered to multiple nodes:
$ ls -l /sys/devices/system/memory/memory21/node*
total 0
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
In that case, the system is able to boot but if later one of theses memory
block is hot-unplugged and then hot-plugged, the sysfs inconsistency is
detected and triggered a BUG_ON():
------------[ cut here ]------------
kernel BUG at /Users/laurent/src/linux-ppc/mm/memory_hotplug.c:1084!
Oops: Exception in kernel mode, sig: 5 [#1]
LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
Modules linked in: rpadlpar_io rpaphp pseries_rng rng_core vmx_crypto gf128mul binfmt_misc ip_tables x_tables xfs libcrc32c crc32c_vpmsum autofs4
CPU: 8 PID: 10256 Comm: drmgr Not tainted 5.9.0-rc1+ #25
NIP: c000000000403f34 LR: c000000000403f2c CTR: 0000000000000000
REGS: c0000004876e3660 TRAP: 0700 Not tainted (5.9.0-rc1+)
MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 24000448 XER: 20040000
CFAR: c000000000846d20 IRQMASK: 0
GPR00: c000000000403f2c c0000004876e38f0 c0000000012f6f00 ffffffffffffffef
GPR04: 0000000000000227 c0000004805ae680 0000000000000000 00000004886f0000
GPR08: 0000000000000226 0000000000000003 0000000000000002 fffffffffffffffd
GPR12: 0000000088000484 c00000001ec96280 0000000000000000 0000000000000000
GPR16: 0000000000000000 0000000000000000 0000000000000004 0000000000000003
GPR20: c00000047814ffe0 c0000007ffff7c08 0000000000000010 c0000000013332c8
GPR24: 0000000000000000 c0000000011f6cc0 0000000000000000 0000000000000000
GPR28: ffffffffffffffef 0000000000000001 0000000150000000 0000000010000000
NIP [c000000000403f34] add_memory_resource+0x244/0x340
LR [c000000000403f2c] add_memory_resource+0x23c/0x340
Call Trace:
[c0000004876e38f0] [c000000000403f2c] add_memory_resource+0x23c/0x340 (unreliable)
[c0000004876e39c0] [c00000000040408c] __add_memory+0x5c/0xf0
[c0000004876e39f0] [c0000000000e2b94] dlpar_add_lmb+0x1b4/0x500
[c0000004876e3ad0] [c0000000000e3888] dlpar_memory+0x1f8/0xb80
[c0000004876e3b60] [c0000000000dc0d0] handle_dlpar_errorlog+0xc0/0x190
[c0000004876e3bd0] [c0000000000dc398] dlpar_store+0x198/0x4a0
[c0000004876e3c90] [c00000000072e630] kobj_attr_store+0x30/0x50
[c0000004876e3cb0] [c00000000051f954] sysfs_kf_write+0x64/0x90
[c0000004876e3cd0] [c00000000051ee40] kernfs_fop_write+0x1b0/0x290
[c0000004876e3d20] [c000000000438dd8] vfs_write+0xe8/0x290
[c0000004876e3d70] [c0000000004391ac] ksys_write+0xdc/0x130
[c0000004876e3dc0] [c000000000034e40] system_call_exception+0x160/0x270
[c0000004876e3e20] [c00000000000d740] system_call_common+0xf0/0x27c
Instruction dump:
48442e35 60000000 0b030000 3cbe0001 7fa3eb78 7bc48402 38a5fffe 7ca5fa14
78a58402 48442db1 60000000 7c7c1b78 <0b030000> 7f23cb78 4bda371d 60000000
---[ end trace 562fd6c109cd0fb2 ]---
This patch addresses the root cause by not relying on the system_state
value to detect whether the call is due to a hot-plug operation. An extra
parameter is needed in register_mem_sect_under_node() detailing whether the
operation is due to a hot-plug operation.
Fixes: 4fbce633910e ("mm/memory_hotplug.c: make register_mem_sect_under_node() a callback of walk_memory_range()")
Signed-off-by: Laurent Dufour <[email protected]>
Cc: [email protected]
Cc: Greg Kroah-Hartman <[email protected]>
Cc: "Rafael J. Wysocki" <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Michal Hocko <[email protected]>
Cc: David Hildenbrand <[email protected]>
---
drivers/base/node.c | 21 ++++++++++++++++-----
include/linux/node.h | 9 ++++++---
mm/memory_hotplug.c | 3 ++-
3 files changed, 24 insertions(+), 9 deletions(-)
diff --git a/drivers/base/node.c b/drivers/base/node.c
index 508b80f6329b..862516c5a5ae 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -762,14 +762,19 @@ static int __ref get_nid_for_pfn(unsigned long pfn)
}
/* register memory section under specified node if it spans that node */
+struct rmsun_args {
+ int nid;
+ enum memplug_context context;
+};
static int register_mem_sect_under_node(struct memory_block *mem_blk,
- void *arg)
+ void *args)
{
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 end_pfn = start_pfn + memory_block_pfns - 1;
- int ret, nid = *(int *)arg;
+ int ret, nid = ((struct rmsun_args *)args)->nid;
unsigned long pfn;
+ enum memplug_context context = ((struct rmsun_args *)args)->context;
for (pfn = start_pfn; pfn <= end_pfn; pfn++) {
int page_nid;
@@ -789,7 +794,7 @@ static int register_mem_sect_under_node(struct memory_block *mem_blk,
* case, during hotplug we know that all pages in the memory
* block belong to the same node.
*/
- if (system_state == SYSTEM_BOOTING) {
+ if (context == MEMPLUG_EARLY) {
page_nid = get_nid_for_pfn(pfn);
if (page_nid < 0)
continue;
@@ -832,10 +837,16 @@ void unregister_memory_block_under_nodes(struct memory_block *mem_blk)
kobject_name(&node_devices[mem_blk->nid]->dev.kobj));
}
-int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn)
+int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
+ enum memplug_context context)
{
+ struct rmsun_args args = {
+ .nid = nid,
+ .context = context,
+ };
+
return walk_memory_blocks(PFN_PHYS(start_pfn),
- PFN_PHYS(end_pfn - start_pfn), (void *)&nid,
+ PFN_PHYS(end_pfn - start_pfn), (void *)&args,
register_mem_sect_under_node);
}
diff --git a/include/linux/node.h b/include/linux/node.h
index 4866f32a02d8..8ff08520488c 100644
--- a/include/linux/node.h
+++ b/include/linux/node.h
@@ -100,10 +100,12 @@ typedef void (*node_registration_func_t)(struct node *);
#if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) && defined(CONFIG_NUMA)
extern int link_mem_sections(int nid, unsigned long start_pfn,
- unsigned long end_pfn);
+ unsigned long end_pfn,
+ enum memplug_context context);
#else
static inline int link_mem_sections(int nid, unsigned long start_pfn,
- unsigned long end_pfn)
+ unsigned long end_pfn,
+ enum memplug_context context)
{
return 0;
}
@@ -128,7 +130,8 @@ static inline int register_one_node(int nid)
if (error)
return error;
/* link memory sections under this node */
- error = link_mem_sections(nid, start_pfn, end_pfn);
+ error = link_mem_sections(nid, start_pfn, end_pfn,
+ MEMPLUG_EARLY);
}
return error;
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index fc21625e42de..912d355ca446 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1080,7 +1080,8 @@ int __ref add_memory_resource(int nid, struct resource *res)
}
/* link memory sections under this node.*/
- ret = link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1));
+ ret = link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1),
+ MEMPLUG_HOTPLUG);
BUG_ON(ret);
/* create new memmap entry */
--
2.28.0
> struct page *start;
> struct memmap_init_callback_data args;
> diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
> index 375515803cd8..cd2bd21d3a4d 100644
> --- a/include/linux/memory_hotplug.h
> +++ b/include/linux/memory_hotplug.h
> @@ -15,6 +15,15 @@ struct memory_block;
> struct resource;
> struct vmem_altmap;
>
> +/*
> + * Memory plugin context, use to differentiate memory added at boot time and
"Memory plugin context" sounds weird.
> + * hot-plugged memory.
> + */
> +enum memplug_context {
> + MEMPLUG_EARLY,
> + MEMPLUG_HOTPLUG,
> +};
Not fully sure this is the right place, though. memory_hotplug.h is
about MEMPLUG_HOTPLUG only ... I'd leave it at the old spot.
--
Thanks,
David / dhildenb
On 11.09.20 18:23, Laurent Dufour wrote:
> Le 11/09/2020 à 16:59, David Hildenbrand a écrit :
>>> struct page *start;
>>> struct memmap_init_callback_data args;
>>> diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
>>> index 375515803cd8..cd2bd21d3a4d 100644
>>> --- a/include/linux/memory_hotplug.h
>>> +++ b/include/linux/memory_hotplug.h
>>> @@ -15,6 +15,15 @@ struct memory_block;
>>> struct resource;
>>> struct vmem_altmap;
>>>
>>> +/*
>>> + * Memory plugin context, use to differentiate memory added at boot time and
>>
>> "Memory plugin context" sounds weird.
>
> Any suggestion ?
>
I'm not a native speaker, but "plugin" sounds weird in this context.
Maybe simply drop it and start right away with "Differentiate memory
added ..."
--
Thanks,
David / dhildenb
On 11.09.20 15:48, Laurent Dufour wrote:
> In register_mem_sect_under_node() the system_state’s value is checked to
> detect whether the operation the call is made during boot time or during an
> hot-plug operation. Unfortunately, that check against SYSTEM_BOOTING is
> wrong because regular memory is registered at SYSTEM_SCHEDULING state. In
> addition memory hot-plug operation can be triggered at this system state
> too by the ACPI. So checking against the system state is not enough.
>
> The consequence is that on system with interleaved node's ranges like this:
> Early memory node ranges
> node 1: [mem 0x0000000000000000-0x000000011fffffff]
> node 2: [mem 0x0000000120000000-0x000000014fffffff]
> node 1: [mem 0x0000000150000000-0x00000001ffffffff]
> node 0: [mem 0x0000000200000000-0x000000048fffffff]
> node 2: [mem 0x0000000490000000-0x00000007ffffffff]
>
> This can be seen on PowerPC LPAR after multiple memory hot-plug and
> hot-unplug operations are done. At the next reboot the node's memory ranges
> can be interleaved and since the call to link_mem_sections() is made in
> topology_init() while the system is in the SYSTEM_SCHEDULING state, the
> node's id is not checked, and the sections registered to multiple nodes:
>
> $ ls -l /sys/devices/system/memory/memory21/node*
> total 0
> lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
> lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
>
> In that case, the system is able to boot but if later one of theses memory
> block is hot-unplugged and then hot-plugged, the sysfs inconsistency is
> detected and triggered a BUG_ON():
>
> ------------[ cut here ]------------
> kernel BUG at /Users/laurent/src/linux-ppc/mm/memory_hotplug.c:1084!
> Oops: Exception in kernel mode, sig: 5 [#1]
> LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
> Modules linked in: rpadlpar_io rpaphp pseries_rng rng_core vmx_crypto gf128mul binfmt_misc ip_tables x_tables xfs libcrc32c crc32c_vpmsum autofs4
> CPU: 8 PID: 10256 Comm: drmgr Not tainted 5.9.0-rc1+ #25
> NIP: c000000000403f34 LR: c000000000403f2c CTR: 0000000000000000
> REGS: c0000004876e3660 TRAP: 0700 Not tainted (5.9.0-rc1+)
> MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 24000448 XER: 20040000
> CFAR: c000000000846d20 IRQMASK: 0
> GPR00: c000000000403f2c c0000004876e38f0 c0000000012f6f00 ffffffffffffffef
> GPR04: 0000000000000227 c0000004805ae680 0000000000000000 00000004886f0000
> GPR08: 0000000000000226 0000000000000003 0000000000000002 fffffffffffffffd
> GPR12: 0000000088000484 c00000001ec96280 0000000000000000 0000000000000000
> GPR16: 0000000000000000 0000000000000000 0000000000000004 0000000000000003
> GPR20: c00000047814ffe0 c0000007ffff7c08 0000000000000010 c0000000013332c8
> GPR24: 0000000000000000 c0000000011f6cc0 0000000000000000 0000000000000000
> GPR28: ffffffffffffffef 0000000000000001 0000000150000000 0000000010000000
> NIP [c000000000403f34] add_memory_resource+0x244/0x340
> LR [c000000000403f2c] add_memory_resource+0x23c/0x340
> Call Trace:
> [c0000004876e38f0] [c000000000403f2c] add_memory_resource+0x23c/0x340 (unreliable)
> [c0000004876e39c0] [c00000000040408c] __add_memory+0x5c/0xf0
> [c0000004876e39f0] [c0000000000e2b94] dlpar_add_lmb+0x1b4/0x500
> [c0000004876e3ad0] [c0000000000e3888] dlpar_memory+0x1f8/0xb80
> [c0000004876e3b60] [c0000000000dc0d0] handle_dlpar_errorlog+0xc0/0x190
> [c0000004876e3bd0] [c0000000000dc398] dlpar_store+0x198/0x4a0
> [c0000004876e3c90] [c00000000072e630] kobj_attr_store+0x30/0x50
> [c0000004876e3cb0] [c00000000051f954] sysfs_kf_write+0x64/0x90
> [c0000004876e3cd0] [c00000000051ee40] kernfs_fop_write+0x1b0/0x290
> [c0000004876e3d20] [c000000000438dd8] vfs_write+0xe8/0x290
> [c0000004876e3d70] [c0000000004391ac] ksys_write+0xdc/0x130
> [c0000004876e3dc0] [c000000000034e40] system_call_exception+0x160/0x270
> [c0000004876e3e20] [c00000000000d740] system_call_common+0xf0/0x27c
> Instruction dump:
> 48442e35 60000000 0b030000 3cbe0001 7fa3eb78 7bc48402 38a5fffe 7ca5fa14
> 78a58402 48442db1 60000000 7c7c1b78 <0b030000> 7f23cb78 4bda371d 60000000
> ---[ end trace 562fd6c109cd0fb2 ]---
>
> This patch addresses the root cause by not relying on the system_state
> value to detect whether the call is due to a hot-plug operation. An extra
> parameter is needed in register_mem_sect_under_node() detailing whether the
> operation is due to a hot-plug operation.
>
> Fixes: 4fbce633910e ("mm/memory_hotplug.c: make register_mem_sect_under_node() a callback of walk_memory_range()")
> Signed-off-by: Laurent Dufour <[email protected]>
> Cc: [email protected]
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: "Rafael J. Wysocki" <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Michal Hocko <[email protected]>
> Cc: David Hildenbrand <[email protected]>
> ---
> drivers/base/node.c | 21 ++++++++++++++++-----
> include/linux/node.h | 9 ++++++---
> mm/memory_hotplug.c | 3 ++-
> 3 files changed, 24 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/base/node.c b/drivers/base/node.c
> index 508b80f6329b..862516c5a5ae 100644
> --- a/drivers/base/node.c
> +++ b/drivers/base/node.c
> @@ -762,14 +762,19 @@ static int __ref get_nid_for_pfn(unsigned long pfn)
> }
>
> /* register memory section under specified node if it spans that node */
> +struct rmsun_args {
> + int nid;
> + enum memplug_context context;
> +};
> static int register_mem_sect_under_node(struct memory_block *mem_blk,
> - void *arg)
> + void *args)
> {
Instead of handling this in register_mem_sect_under_node(), I
think it would be better two have two separate
register_mem_sect_under_node() implementations.
static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
void *arg)
{
const int nid = *(int *)arg;
int ret;
/* Hotplugged memory has no holes and belongs to a single node. */
mem_blk->nid = nid;
ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
&mem_blk->dev.kobj,
kobject_name(&mem_blk->dev.kobj));
if (ret)
returnr et;
return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
&node_devices[nid]->dev.kobj,
kobject_name(&node_devices[nid]->dev.kobj));
}
Cleaner, right? :) No unnecessary checks.
One could argue if link_mem_section_hotplug() would be better than passing around the context.
--
Thanks,
David / dhildenb
Le 14/09/2020 à 09:57, David Hildenbrand a écrit :
> On 11.09.20 15:48, Laurent Dufour wrote:
>> In register_mem_sect_under_node() the system_state’s value is checked to
>> detect whether the operation the call is made during boot time or during an
>> hot-plug operation. Unfortunately, that check against SYSTEM_BOOTING is
>> wrong because regular memory is registered at SYSTEM_SCHEDULING state. In
>> addition memory hot-plug operation can be triggered at this system state
>> too by the ACPI. So checking against the system state is not enough.
>>
>> The consequence is that on system with interleaved node's ranges like this:
>> Early memory node ranges
>> node 1: [mem 0x0000000000000000-0x000000011fffffff]
>> node 2: [mem 0x0000000120000000-0x000000014fffffff]
>> node 1: [mem 0x0000000150000000-0x00000001ffffffff]
>> node 0: [mem 0x0000000200000000-0x000000048fffffff]
>> node 2: [mem 0x0000000490000000-0x00000007ffffffff]
>>
>> This can be seen on PowerPC LPAR after multiple memory hot-plug and
>> hot-unplug operations are done. At the next reboot the node's memory ranges
>> can be interleaved and since the call to link_mem_sections() is made in
>> topology_init() while the system is in the SYSTEM_SCHEDULING state, the
>> node's id is not checked, and the sections registered to multiple nodes:
>>
>> $ ls -l /sys/devices/system/memory/memory21/node*
>> total 0
>> lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
>> lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
>>
>> In that case, the system is able to boot but if later one of theses memory
>> block is hot-unplugged and then hot-plugged, the sysfs inconsistency is
>> detected and triggered a BUG_ON():
>>
>> ------------[ cut here ]------------
>> kernel BUG at /Users/laurent/src/linux-ppc/mm/memory_hotplug.c:1084!
>> Oops: Exception in kernel mode, sig: 5 [#1]
>> LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
>> Modules linked in: rpadlpar_io rpaphp pseries_rng rng_core vmx_crypto gf128mul binfmt_misc ip_tables x_tables xfs libcrc32c crc32c_vpmsum autofs4
>> CPU: 8 PID: 10256 Comm: drmgr Not tainted 5.9.0-rc1+ #25
>> NIP: c000000000403f34 LR: c000000000403f2c CTR: 0000000000000000
>> REGS: c0000004876e3660 TRAP: 0700 Not tainted (5.9.0-rc1+)
>> MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 24000448 XER: 20040000
>> CFAR: c000000000846d20 IRQMASK: 0
>> GPR00: c000000000403f2c c0000004876e38f0 c0000000012f6f00 ffffffffffffffef
>> GPR04: 0000000000000227 c0000004805ae680 0000000000000000 00000004886f0000
>> GPR08: 0000000000000226 0000000000000003 0000000000000002 fffffffffffffffd
>> GPR12: 0000000088000484 c00000001ec96280 0000000000000000 0000000000000000
>> GPR16: 0000000000000000 0000000000000000 0000000000000004 0000000000000003
>> GPR20: c00000047814ffe0 c0000007ffff7c08 0000000000000010 c0000000013332c8
>> GPR24: 0000000000000000 c0000000011f6cc0 0000000000000000 0000000000000000
>> GPR28: ffffffffffffffef 0000000000000001 0000000150000000 0000000010000000
>> NIP [c000000000403f34] add_memory_resource+0x244/0x340
>> LR [c000000000403f2c] add_memory_resource+0x23c/0x340
>> Call Trace:
>> [c0000004876e38f0] [c000000000403f2c] add_memory_resource+0x23c/0x340 (unreliable)
>> [c0000004876e39c0] [c00000000040408c] __add_memory+0x5c/0xf0
>> [c0000004876e39f0] [c0000000000e2b94] dlpar_add_lmb+0x1b4/0x500
>> [c0000004876e3ad0] [c0000000000e3888] dlpar_memory+0x1f8/0xb80
>> [c0000004876e3b60] [c0000000000dc0d0] handle_dlpar_errorlog+0xc0/0x190
>> [c0000004876e3bd0] [c0000000000dc398] dlpar_store+0x198/0x4a0
>> [c0000004876e3c90] [c00000000072e630] kobj_attr_store+0x30/0x50
>> [c0000004876e3cb0] [c00000000051f954] sysfs_kf_write+0x64/0x90
>> [c0000004876e3cd0] [c00000000051ee40] kernfs_fop_write+0x1b0/0x290
>> [c0000004876e3d20] [c000000000438dd8] vfs_write+0xe8/0x290
>> [c0000004876e3d70] [c0000000004391ac] ksys_write+0xdc/0x130
>> [c0000004876e3dc0] [c000000000034e40] system_call_exception+0x160/0x270
>> [c0000004876e3e20] [c00000000000d740] system_call_common+0xf0/0x27c
>> Instruction dump:
>> 48442e35 60000000 0b030000 3cbe0001 7fa3eb78 7bc48402 38a5fffe 7ca5fa14
>> 78a58402 48442db1 60000000 7c7c1b78 <0b030000> 7f23cb78 4bda371d 60000000
>> ---[ end trace 562fd6c109cd0fb2 ]---
>>
>> This patch addresses the root cause by not relying on the system_state
>> value to detect whether the call is due to a hot-plug operation. An extra
>> parameter is needed in register_mem_sect_under_node() detailing whether the
>> operation is due to a hot-plug operation.
>>
>> Fixes: 4fbce633910e ("mm/memory_hotplug.c: make register_mem_sect_under_node() a callback of walk_memory_range()")
>> Signed-off-by: Laurent Dufour <[email protected]>
>> Cc: [email protected]
>> Cc: Greg Kroah-Hartman <[email protected]>
>> Cc: "Rafael J. Wysocki" <[email protected]>
>> Cc: Andrew Morton <[email protected]>
>> Cc: Michal Hocko <[email protected]>
>> Cc: David Hildenbrand <[email protected]>
>> ---
>> drivers/base/node.c | 21 ++++++++++++++++-----
>> include/linux/node.h | 9 ++++++---
>> mm/memory_hotplug.c | 3 ++-
>> 3 files changed, 24 insertions(+), 9 deletions(-)
>>
>> diff --git a/drivers/base/node.c b/drivers/base/node.c
>> index 508b80f6329b..862516c5a5ae 100644
>> --- a/drivers/base/node.c
>> +++ b/drivers/base/node.c
>> @@ -762,14 +762,19 @@ static int __ref get_nid_for_pfn(unsigned long pfn)
>> }
>>
>> /* register memory section under specified node if it spans that node */
>> +struct rmsun_args {
>> + int nid;
>> + enum memplug_context context;
>> +};
>> static int register_mem_sect_under_node(struct memory_block *mem_blk,
>> - void *arg)
>> + void *args)
>> {
>
> Instead of handling this in register_mem_sect_under_node(), I
> think it would be better two have two separate
> register_mem_sect_under_node() implementations.
>
> static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
> void *arg)
> {
> const int nid = *(int *)arg;
> int ret;
>
> /* Hotplugged memory has no holes and belongs to a single node. */
> mem_blk->nid = nid;
> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
> &mem_blk->dev.kobj,
> kobject_name(&mem_blk->dev.kobj));
> if (ret)
> returnr et;
> return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
> &node_devices[nid]->dev.kobj,
> kobject_name(&node_devices[nid]->dev.kobj));
>
> }
>
> Cleaner, right? :) No unnecessary checks.
Cela me laisse perplexe. Dans un sens, cela simplifie le patch, mais cela
revient à dupliquer le code pour ne supprimer qu'une seule vérification.
I'm fine with the both options, Ijust want to be sure that all agree on the
direction before changing my series in that way.
>
> One could argue if link_mem_section_hotplug() would be better than passing around the context.
>
I do ;)
On Mon, Sep 14, 2020 at 09:57:46AM +0200, David Hildenbrand wrote:
> > /* register memory section under specified node if it spans that node */
> > +struct rmsun_args {
> > + int nid;
> > + enum memplug_context context;
> > +};
Uhmf, that is a not so descriptive name.
> Instead of handling this in register_mem_sect_under_node(), I
> think it would be better two have two separate
> register_mem_sect_under_node() implementations.
>
> static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
> void *arg)
> {
> const int nid = *(int *)arg;
> int ret;
>
> /* Hotplugged memory has no holes and belongs to a single node. */
> mem_blk->nid = nid;
> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
> &mem_blk->dev.kobj,
> kobject_name(&mem_blk->dev.kobj));
> if (ret)
> returnr et;
> return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
> &node_devices[nid]->dev.kobj,
> kobject_name(&node_devices[nid]->dev.kobj));
>
> }
>
> Cleaner, right? :) No unnecessary checks.
I tend to agree here, I like more a simplistic version for hotplug.
> One could argue if link_mem_section_hotplug() would be better than passing around the context.
I am not sure if I would duplicate the code there.
We could just pass the pointer of the function we want to call to
link_mem_sections? either register_mem_sect_under_node_hotplug or
register_mem_sect_under_node_early?
Would not that be clean and clear enough?
--
Oscar Salvador
SUSE L3
>> static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
>> void *arg)
>> {
>> const int nid = *(int *)arg;
>> int ret;
>>
>> /* Hotplugged memory has no holes and belongs to a single node. */
>> mem_blk->nid = nid;
>> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
>> &mem_blk->dev.kobj,
>> kobject_name(&mem_blk->dev.kobj));
>> if (ret)
>> returnr et;
>> return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
>> &node_devices[nid]->dev.kobj,
>> kobject_name(&node_devices[nid]->dev.kobj));
>>
>> }
>>
>> Cleaner, right? :) No unnecessary checks.
>
> I tend to agree here, I like more a simplistic version for hotplug.
>
... and while we're at it, we should rename register_mem_sect_under_node
to something like "register_memory_block_under_node" - "section" is a
legacy leftover here.
We could factor out both sysfs_create_link_nowarn() calls into something
like "do_register_memory_block_under_node" or similar, to minimize code
duplication.
>> One could argue if link_mem_section_hotplug() would be better than passing around the context.
>
> I am not sure if I would duplicate the code there.
> We could just pass the pointer of the function we want to call to
> link_mem_sections? either register_mem_sect_under_node_hotplug or
> register_mem_sect_under_node_early?
> Would not that be clean and clear enough?
I don't particularly like passing around function pointers where it can
be avoided (e.g., here exporting 3 functions now instead 1). Makes the
interface harder to get IMHO. But I don't really care about that
interface, easy to change later on.
--
Thanks,
David / dhildenb
Le 14/09/2020 à 10:19, Oscar Salvador a écrit :
> On Mon, Sep 14, 2020 at 09:57:46AM +0200, David Hildenbrand wrote:
>>> /* register memory section under specified node if it spans that node */
>>> +struct rmsun_args {
>>> + int nid;
>>> + enum memplug_context context;
>>> +};
>
> Uhmf, that is a not so descriptive name.
I do agree, but didn't have a better idea.
Anyway this will disappear since the choosen direction is to have 2 callbacks.
>
>> Instead of handling this in register_mem_sect_under_node(), I
>> think it would be better two have two separate
>> register_mem_sect_under_node() implementations.
>>
>> static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
>> void *arg)
>> {
>> const int nid = *(int *)arg;
>> int ret;
>>
>> /* Hotplugged memory has no holes and belongs to a single node. */
>> mem_blk->nid = nid;
>> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
>> &mem_blk->dev.kobj,
>> kobject_name(&mem_blk->dev.kobj));
>> if (ret)
>> returnr et;
>> return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
>> &node_devices[nid]->dev.kobj,
>> kobject_name(&node_devices[nid]->dev.kobj));
>>
>> }
>>
>> Cleaner, right? :) No unnecessary checks.
>
> I tend to agree here, I like more a simplistic version for hotplug.
>
>> One could argue if link_mem_section_hotplug() would be better than passing around the context.
>
> I am not sure if I would duplicate the code there.
> We could just pass the pointer of the function we want to call to
> link_mem_sections? either register_mem_sect_under_node_hotplug or
> register_mem_sect_under_node_early?
> Would not that be clean and clear enough?
That would expose the register_mem_sect_under_node*() prototype to the caller.
I'm wondering if that would be cleaner than passing a MEMPLUG_* constant value
to link_mem_sections() and let it choose the right callback.
On Fri 11-09-20 15:48:29, Laurent Dufour wrote:
> The memmap_context is used to detect whether a memory operation is due to a
> hot-add operation or happening at boot time.
>
> Makes it general to the hotplug operation, renaming it at memplug_context
> and move its define in the corresponding header.
>
> There is no functional change introduced by this patch
I do not want to nit picking on naming but we want to look at this from
the initialization POV rather than hotplug. So....
>
> Suggested-by: David Hildenbrand <[email protected]>
> Signed-off-by: Laurent Dufour <[email protected]>
[...]
> diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
> index 375515803cd8..cd2bd21d3a4d 100644
> --- a/include/linux/memory_hotplug.h
> +++ b/include/linux/memory_hotplug.h
> @@ -15,6 +15,15 @@ struct memory_block;
> struct resource;
> struct vmem_altmap;
>
> +/*
> + * Memory plugin context, use to differentiate memory added at boot time and
> + * hot-plugged memory.
> + */
> +enum memplug_context {
> + MEMPLUG_EARLY,
> + MEMPLUG_HOTPLUG,
> +};
/*
* Memory initialization context, use to differentiate memory added by
* the platform statically or via memory hotplug interface.
*/
enum meminit_context {
MEMINIT_EARLY,
MEMINIT_HOTPLUG
}
--
Michal Hocko
SUSE Labs
Le 14/09/2020 à 10:49, Michal Hocko a écrit :
> On Fri 11-09-20 15:48:29, Laurent Dufour wrote:
>> The memmap_context is used to detect whether a memory operation is due to a
>> hot-add operation or happening at boot time.
>>
>> Makes it general to the hotplug operation, renaming it at memplug_context
>> and move its define in the corresponding header.
>>
>> There is no functional change introduced by this patch
>
> I do not want to nit picking on naming but we want to look at this from
> the initialization POV rather than hotplug. So....
>>
>> Suggested-by: David Hildenbrand <[email protected]>
>> Signed-off-by: Laurent Dufour <[email protected]>
> [...]
>> diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h
>> index 375515803cd8..cd2bd21d3a4d 100644
>> --- a/include/linux/memory_hotplug.h
>> +++ b/include/linux/memory_hotplug.h
>> @@ -15,6 +15,15 @@ struct memory_block;
>> struct resource;
>> struct vmem_altmap;
>>
>> +/*
>> + * Memory plugin context, use to differentiate memory added at boot time and
>> + * hot-plugged memory.
>> + */
>> +enum memplug_context {
>> + MEMPLUG_EARLY,
>> + MEMPLUG_HOTPLUG,
>> +};
>
> /*
> * Memory initialization context, use to differentiate memory added by
> * the platform statically or via memory hotplug interface.
> */
> enum meminit_context {
> MEMINIT_EARLY,
> MEMINIT_HOTPLUG
> }
>
Sounds good too.
What about its definition's place, in include/mm/zone.h as David suggested ?
On Fri 11-09-20 15:48:30, Laurent Dufour wrote:
> In register_mem_sect_under_node() the system_state’s value is checked to
> detect whether the operation the call is made during boot time or during an
> hot-plug operation. Unfortunately, that check against SYSTEM_BOOTING is
> wrong because regular memory is registered at SYSTEM_SCHEDULING state. In
> addition memory hot-plug operation can be triggered at this system state
> too by the ACPI.
Please add
(an example would be Qemu to replay hot added memory after reboot).
> So checking against the system state is not enough.
>
> The consequence is that on system with interleaved node's ranges like this:
> Early memory node ranges
> node 1: [mem 0x0000000000000000-0x000000011fffffff]
> node 2: [mem 0x0000000120000000-0x000000014fffffff]
> node 1: [mem 0x0000000150000000-0x00000001ffffffff]
> node 0: [mem 0x0000000200000000-0x000000048fffffff]
> node 2: [mem 0x0000000490000000-0x00000007ffffffff]
>
> This can be seen on PowerPC LPAR after multiple memory hot-plug and
> hot-unplug operations are done. At the next reboot the node's memory ranges
> can be interleaved and since the call to link_mem_sections() is made in
> topology_init() while the system is in the SYSTEM_SCHEDULING state, the
> node's id is not checked, and the sections registered to multiple nodes:
>
> $ ls -l /sys/devices/system/memory/memory21/node*
> total 0
> lrwxrwxrwx 1 root root 0 Aug 24 05:27 node1 -> ../../node/node1
> lrwxrwxrwx 1 root root 0 Aug 24 05:27 node2 -> ../../node/node2
>
> In that case, the system is able to boot but if later one of theses memory
> block is hot-unplugged and then hot-plugged, the sysfs inconsistency is
> detected and triggered a BUG_ON():
>
> ------------[ cut here ]------------
> kernel BUG at /Users/laurent/src/linux-ppc/mm/memory_hotplug.c:1084!
> Oops: Exception in kernel mode, sig: 5 [#1]
> LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
> Modules linked in: rpadlpar_io rpaphp pseries_rng rng_core vmx_crypto gf128mul binfmt_misc ip_tables x_tables xfs libcrc32c crc32c_vpmsum autofs4
> CPU: 8 PID: 10256 Comm: drmgr Not tainted 5.9.0-rc1+ #25
> NIP: c000000000403f34 LR: c000000000403f2c CTR: 0000000000000000
> REGS: c0000004876e3660 TRAP: 0700 Not tainted (5.9.0-rc1+)
> MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 24000448 XER: 20040000
> CFAR: c000000000846d20 IRQMASK: 0
> GPR00: c000000000403f2c c0000004876e38f0 c0000000012f6f00 ffffffffffffffef
> GPR04: 0000000000000227 c0000004805ae680 0000000000000000 00000004886f0000
> GPR08: 0000000000000226 0000000000000003 0000000000000002 fffffffffffffffd
> GPR12: 0000000088000484 c00000001ec96280 0000000000000000 0000000000000000
> GPR16: 0000000000000000 0000000000000000 0000000000000004 0000000000000003
> GPR20: c00000047814ffe0 c0000007ffff7c08 0000000000000010 c0000000013332c8
> GPR24: 0000000000000000 c0000000011f6cc0 0000000000000000 0000000000000000
> GPR28: ffffffffffffffef 0000000000000001 0000000150000000 0000000010000000
> NIP [c000000000403f34] add_memory_resource+0x244/0x340
> LR [c000000000403f2c] add_memory_resource+0x23c/0x340
> Call Trace:
> [c0000004876e38f0] [c000000000403f2c] add_memory_resource+0x23c/0x340 (unreliable)
> [c0000004876e39c0] [c00000000040408c] __add_memory+0x5c/0xf0
> [c0000004876e39f0] [c0000000000e2b94] dlpar_add_lmb+0x1b4/0x500
> [c0000004876e3ad0] [c0000000000e3888] dlpar_memory+0x1f8/0xb80
> [c0000004876e3b60] [c0000000000dc0d0] handle_dlpar_errorlog+0xc0/0x190
> [c0000004876e3bd0] [c0000000000dc398] dlpar_store+0x198/0x4a0
> [c0000004876e3c90] [c00000000072e630] kobj_attr_store+0x30/0x50
> [c0000004876e3cb0] [c00000000051f954] sysfs_kf_write+0x64/0x90
> [c0000004876e3cd0] [c00000000051ee40] kernfs_fop_write+0x1b0/0x290
> [c0000004876e3d20] [c000000000438dd8] vfs_write+0xe8/0x290
> [c0000004876e3d70] [c0000000004391ac] ksys_write+0xdc/0x130
> [c0000004876e3dc0] [c000000000034e40] system_call_exception+0x160/0x270
> [c0000004876e3e20] [c00000000000d740] system_call_common+0xf0/0x27c
> Instruction dump:
> 48442e35 60000000 0b030000 3cbe0001 7fa3eb78 7bc48402 38a5fffe 7ca5fa14
> 78a58402 48442db1 60000000 7c7c1b78 <0b030000> 7f23cb78 4bda371d 60000000
> ---[ end trace 562fd6c109cd0fb2 ]---
>
> This patch addresses the root cause by not relying on the system_state
> value to detect whether the call is due to a hot-plug operation. An extra
> parameter is needed in register_mem_sect_under_node() detailing whether the
> operation is due to a hot-plug operation.
Thanks this description is much more clear!
> Fixes: 4fbce633910e ("mm/memory_hotplug.c: make register_mem_sect_under_node() a callback of walk_memory_range()")
> Signed-off-by: Laurent Dufour <[email protected]>
> Cc: [email protected]
> Cc: Greg Kroah-Hartman <[email protected]>
> Cc: "Rafael J. Wysocki" <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Michal Hocko <[email protected]>
> Cc: David Hildenbrand <[email protected]>
Wrt to the specific implementation I cannot say I would care much.
Duplicating the code will make the two callers simpler but there will be
a code duplication. We have that weird intermix for hotplug vs. early
init at other places. So I would be very lightly prefering this patch
and do a larger cleanup on top of that. It is at least clear what the
patch does this way. So
Acked-by: Michal Hocko <[email protected]>
> ---
> drivers/base/node.c | 21 ++++++++++++++++-----
> include/linux/node.h | 9 ++++++---
> mm/memory_hotplug.c | 3 ++-
> 3 files changed, 24 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/base/node.c b/drivers/base/node.c
> index 508b80f6329b..862516c5a5ae 100644
> --- a/drivers/base/node.c
> +++ b/drivers/base/node.c
> @@ -762,14 +762,19 @@ static int __ref get_nid_for_pfn(unsigned long pfn)
> }
>
> /* register memory section under specified node if it spans that node */
> +struct rmsun_args {
> + int nid;
> + enum memplug_context context;
> +};
> static int register_mem_sect_under_node(struct memory_block *mem_blk,
> - void *arg)
> + void *args)
> {
> 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 end_pfn = start_pfn + memory_block_pfns - 1;
> - int ret, nid = *(int *)arg;
> + int ret, nid = ((struct rmsun_args *)args)->nid;
> unsigned long pfn;
> + enum memplug_context context = ((struct rmsun_args *)args)->context;
>
> for (pfn = start_pfn; pfn <= end_pfn; pfn++) {
> int page_nid;
> @@ -789,7 +794,7 @@ static int register_mem_sect_under_node(struct memory_block *mem_blk,
> * case, during hotplug we know that all pages in the memory
> * block belong to the same node.
> */
> - if (system_state == SYSTEM_BOOTING) {
> + if (context == MEMPLUG_EARLY) {
> page_nid = get_nid_for_pfn(pfn);
> if (page_nid < 0)
> continue;
> @@ -832,10 +837,16 @@ void unregister_memory_block_under_nodes(struct memory_block *mem_blk)
> kobject_name(&node_devices[mem_blk->nid]->dev.kobj));
> }
>
> -int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn)
> +int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
> + enum memplug_context context)
> {
> + struct rmsun_args args = {
> + .nid = nid,
> + .context = context,
> + };
> +
> return walk_memory_blocks(PFN_PHYS(start_pfn),
> - PFN_PHYS(end_pfn - start_pfn), (void *)&nid,
> + PFN_PHYS(end_pfn - start_pfn), (void *)&args,
> register_mem_sect_under_node);
> }
>
> diff --git a/include/linux/node.h b/include/linux/node.h
> index 4866f32a02d8..8ff08520488c 100644
> --- a/include/linux/node.h
> +++ b/include/linux/node.h
> @@ -100,10 +100,12 @@ typedef void (*node_registration_func_t)(struct node *);
>
> #if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) && defined(CONFIG_NUMA)
> extern int link_mem_sections(int nid, unsigned long start_pfn,
> - unsigned long end_pfn);
> + unsigned long end_pfn,
> + enum memplug_context context);
> #else
> static inline int link_mem_sections(int nid, unsigned long start_pfn,
> - unsigned long end_pfn)
> + unsigned long end_pfn,
> + enum memplug_context context)
> {
> return 0;
> }
> @@ -128,7 +130,8 @@ static inline int register_one_node(int nid)
> if (error)
> return error;
> /* link memory sections under this node */
> - error = link_mem_sections(nid, start_pfn, end_pfn);
> + error = link_mem_sections(nid, start_pfn, end_pfn,
> + MEMPLUG_EARLY);
> }
>
> return error;
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index fc21625e42de..912d355ca446 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1080,7 +1080,8 @@ int __ref add_memory_resource(int nid, struct resource *res)
> }
>
> /* link memory sections under this node.*/
> - ret = link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1));
> + ret = link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1),
> + MEMPLUG_HOTPLUG);
> BUG_ON(ret);
>
> /* create new memmap entry */
> --
> 2.28.0
--
Michal Hocko
SUSE Labs
On Fri 11-09-20 15:48:31, Laurent Dufour wrote:
> At boot time, or when doing memory hot-add operations, if the links in
> sysfs can't be created, the system is still able to run, so just report the
> error in the kernel log.
.. rather than BUG_ON and potentially make system unusable because the
callpath can be called with locks held etc...
> Since the number of memory blocks managed could be high, the messages are
> rate limited.
>
> As a consequence, link_mem_sections() has no status to report anymore.
>
> Signed-off-by: Laurent Dufour <[email protected]>
> Cc: David Hildenbrand <[email protected]>
Acked-by: Michal Hocko <[email protected]>
Thanks!
> ---
> drivers/base/node.c | 25 +++++++++++++++++--------
> include/linux/node.h | 17 ++++++++---------
> mm/memory_hotplug.c | 5 ++---
> 3 files changed, 27 insertions(+), 20 deletions(-)
>
> diff --git a/drivers/base/node.c b/drivers/base/node.c
> index 862516c5a5ae..749a1c8ea992 100644
> --- a/drivers/base/node.c
> +++ b/drivers/base/node.c
> @@ -811,12 +811,21 @@ static int register_mem_sect_under_node(struct memory_block *mem_blk,
> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
> &mem_blk->dev.kobj,
> kobject_name(&mem_blk->dev.kobj));
> - if (ret)
> - return ret;
> + if (ret && ret != -EEXIST)
> + pr_err_ratelimited(
> + "can't create %s to %s link in sysfs (%d)\n",
> + kobject_name(&node_devices[nid]->dev.kobj),
> + kobject_name(&mem_blk->dev.kobj), ret);
>
> - return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
> + ret = sysfs_create_link_nowarn(&mem_blk->dev.kobj,
> &node_devices[nid]->dev.kobj,
> kobject_name(&node_devices[nid]->dev.kobj));
> + if (ret && ret != -EEXIST)
> + pr_err_ratelimited(
> + "can't create %s to %s link in sysfs (%d)\n",
> + kobject_name(&mem_blk->dev.kobj),
> + kobject_name(&node_devices[nid]->dev.kobj),
> + ret);
> }
> /* mem section does not span the specified node */
> return 0;
> @@ -837,17 +846,17 @@ void unregister_memory_block_under_nodes(struct memory_block *mem_blk)
> kobject_name(&node_devices[mem_blk->nid]->dev.kobj));
> }
>
> -int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
> - enum memplug_context context)
> +void link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
> + enum memplug_context context)
> {
> struct rmsun_args args = {
> .nid = nid,
> .context = context,
> };
>
> - return walk_memory_blocks(PFN_PHYS(start_pfn),
> - PFN_PHYS(end_pfn - start_pfn), (void *)&args,
> - register_mem_sect_under_node);
> + walk_memory_blocks(PFN_PHYS(start_pfn),
> + PFN_PHYS(end_pfn - start_pfn), (void *)&args,
> + register_mem_sect_under_node);
> }
>
> #ifdef CONFIG_HUGETLBFS
> diff --git a/include/linux/node.h b/include/linux/node.h
> index 8ff08520488c..6bdd6f3ed3aa 100644
> --- a/include/linux/node.h
> +++ b/include/linux/node.h
> @@ -99,15 +99,14 @@ extern struct node *node_devices[];
> typedef void (*node_registration_func_t)(struct node *);
>
> #if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) && defined(CONFIG_NUMA)
> -extern int link_mem_sections(int nid, unsigned long start_pfn,
> - unsigned long end_pfn,
> - enum memplug_context context);
> +void link_mem_sections(int nid, unsigned long start_pfn,
> + unsigned long end_pfn,
> + enum memplug_context context);
> #else
> -static inline int link_mem_sections(int nid, unsigned long start_pfn,
> - unsigned long end_pfn,
> - enum memplug_context context)
> +static inline void link_mem_sections(int nid, unsigned long start_pfn,
> + unsigned long end_pfn,
> + enum memplug_context context)
> {
> - return 0;
> }
> #endif
>
> @@ -130,8 +129,8 @@ static inline int register_one_node(int nid)
> if (error)
> return error;
> /* link memory sections under this node */
> - error = link_mem_sections(nid, start_pfn, end_pfn,
> - MEMPLUG_EARLY);
> + link_mem_sections(nid, start_pfn, end_pfn,
> + MEMPLUG_EARLY);
> }
>
> return error;
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 912d355ca446..668418071a49 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -1080,9 +1080,8 @@ int __ref add_memory_resource(int nid, struct resource *res)
> }
>
> /* link memory sections under this node.*/
> - ret = link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1),
> - MEMPLUG_HOTPLUG);
> - BUG_ON(ret);
> + link_mem_sections(nid, PFN_DOWN(start), PFN_UP(start + size - 1),
> + MEMPLUG_HOTPLUG);
>
> /* create new memmap entry */
> if (!strcmp(res->name, "System RAM"))
> --
> 2.28.0
--
Michal Hocko
SUSE Labs
On Mon 14-09-20 10:51:06, Laurent Dufour wrote:
> Le 14/09/2020 ? 10:49, Michal Hocko a ?crit?:
[...]
> > /*
> > * Memory initialization context, use to differentiate memory added by
> > * the platform statically or via memory hotplug interface.
> > */
> > enum meminit_context {
> > MEMINIT_EARLY,
> > MEMINIT_HOTPLUG
> > }
> >
>
> Sounds good too.
> What about its definition's place, in include/mm/zone.h as David suggested ?
Yes. This shouldn't be in the hotplug proper.
--
Michal Hocko
SUSE Labs
Le 14/09/2020 à 10:31, David Hildenbrand a écrit :
>>> static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
>>> void *arg)
>>> {
>>> const int nid = *(int *)arg;
>>> int ret;
>>>
>>> /* Hotplugged memory has no holes and belongs to a single node. */
>>> mem_blk->nid = nid;
>>> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
>>> &mem_blk->dev.kobj,
>>> kobject_name(&mem_blk->dev.kobj));
>>> if (ret)
>>> returnr et;
>>> return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
>>> &node_devices[nid]->dev.kobj,
>>> kobject_name(&node_devices[nid]->dev.kobj));
>>>
>>> }
>>>
>>> Cleaner, right? :) No unnecessary checks.
>>
>> I tend to agree here, I like more a simplistic version for hotplug.
>>
>
> ... and while we're at it, we should rename register_mem_sect_under_node
> to something like "register_memory_block_under_node" - "section" is a
> legacy leftover here.
>
> We could factor out both sysfs_create_link_nowarn() calls into something
> like "do_register_memory_block_under_node" or similar, to minimize code
> duplication.
>
>>> One could argue if link_mem_section_hotplug() would be better than passing around the context.
>>
>> I am not sure if I would duplicate the code there.
>> We could just pass the pointer of the function we want to call to
>> link_mem_sections? either register_mem_sect_under_node_hotplug or
>> register_mem_sect_under_node_early?
>> Would not that be clean and clear enough?
>
> I don't particularly like passing around function pointers where it can
> be avoided (e.g., here exporting 3 functions now instead 1). Makes the
> interface harder to get IMHO. But I don't really care about that
> interface, easy to change later on.
>
This would lead to the following.
Do everyone agree?
diff --git a/drivers/base/node.c b/drivers/base/node.c
index 508b80f6329b..444808a7c9b6 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -761,9 +761,32 @@ static int __ref get_nid_for_pfn(unsigned long pfn)
return pfn_to_nid(pfn);
}
+static int do_register_memory_block_under_node(int nid,
+ struct memory_block *mem_blk)
+{
+ int ret;
+
+ /*
+ * If this memory block spans multiple nodes, we only indicate
+ * the last processed node.
+ */
+ mem_blk->nid = nid;
+
+ ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
+ &mem_blk->dev.kobj,
+ kobject_name(&mem_blk->dev.kobj));
+ if (ret)
+ return ret;
+
+ return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
+ &node_devices[nid]->dev.kobj,
+ kobject_name(&node_devices[nid]->dev.kobj));
+
+}
+
/* register memory section under specified node if it spans that node */
-static int register_mem_sect_under_node(struct memory_block *mem_blk,
- void *arg)
+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);
@@ -785,38 +808,35 @@ static int register_mem_sect_under_node(struct
memory_block *mem_blk,
}
/*
- * We need to check if page belongs to nid only for the boot
- * case, during hotplug we know that all pages in the memory
- * block belong to the same node.
+ * We need to check if page belongs to nid only at the boot
+ * case because node's ranges can be interleaved.
*/
- if (system_state == SYSTEM_BOOTING) {
- page_nid = get_nid_for_pfn(pfn);
- if (page_nid < 0)
- continue;
- if (page_nid != nid)
- continue;
- }
-
- /*
- * If this memory block spans multiple nodes, we only indicate
- * the last processed node.
- */
- mem_blk->nid = nid;
+ page_nid = get_nid_for_pfn(pfn);
+ if (page_nid < 0)
+ continue;
+ if (page_nid != nid)
+ continue;
- ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
- &mem_blk->dev.kobj,
- kobject_name(&mem_blk->dev.kobj));
+ ret = do_register_memory_block_under_node(nid, mem_blk);
if (ret)
return ret;
-
- return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
- &node_devices[nid]->dev.kobj,
- kobject_name(&node_devices[nid]->dev.kobj));
}
/* mem section does not span the specified node */
return 0;
}
+/*
+ * During hotplug we know that all pages in the memory block belong to the same
+ * node.
+ */
+static int register_mem_block_under_node_hotplug(struct memory_block *mem_blk,
+ void *arg)
+{
+ int nid = *(int *)arg;
+
+ return do_register_memory_block_under_node(nid, mem_blk);
+}
+
/*
* Unregister a memory block device under the node it spans. Memory blocks
* with multiple nodes cannot be offlined and therefore also never be removed.
@@ -832,11 +852,19 @@ void unregister_memory_block_under_nodes(struct
memory_block *mem_blk)
kobject_name(&node_devices[mem_blk->nid]->dev.kobj));
}
-int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn)
+int link_mem_sections(int nid, unsigned long start_pfn, unsigned long end_pfn,
+ enum memplug_context context)
{
+ walk_memory_blocks_func_t func;
+
+ if (context == MEMPLUG_HOTPLUG)
+ func = register_mem_block_under_node_hotplug;
+ else
+ func = register_mem_block_under_node_early;
+
return walk_memory_blocks(PFN_PHYS(start_pfn),
PFN_PHYS(end_pfn - start_pfn), (void *)&nid,
- register_mem_sect_under_node);
+ func);
}
#ifdef CONFIG_HUGETLBFS
On 14.09.20 11:16, Laurent Dufour wrote:
> Le 14/09/2020 à 10:31, David Hildenbrand a écrit :
>>>> static int register_mem_sect_under_node_hotplug(struct memory_block *mem_blk,
>>>> void *arg)
>>>> {
>>>> const int nid = *(int *)arg;
>>>> int ret;
>>>>
>>>> /* Hotplugged memory has no holes and belongs to a single node. */
>>>> mem_blk->nid = nid;
>>>> ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
>>>> &mem_blk->dev.kobj,
>>>> kobject_name(&mem_blk->dev.kobj));
>>>> if (ret)
>>>> returnr et;
>>>> return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
>>>> &node_devices[nid]->dev.kobj,
>>>> kobject_name(&node_devices[nid]->dev.kobj));
>>>>
>>>> }
>>>>
>>>> Cleaner, right? :) No unnecessary checks.
>>>
>>> I tend to agree here, I like more a simplistic version for hotplug.
>>>
>>
>> ... and while we're at it, we should rename register_mem_sect_under_node
>> to something like "register_memory_block_under_node" - "section" is a
>> legacy leftover here.
>>
>> We could factor out both sysfs_create_link_nowarn() calls into something
>> like "do_register_memory_block_under_node" or similar, to minimize code
>> duplication.
>>
>>>> One could argue if link_mem_section_hotplug() would be better than passing around the context.
>>>
>>> I am not sure if I would duplicate the code there.
>>> We could just pass the pointer of the function we want to call to
>>> link_mem_sections? either register_mem_sect_under_node_hotplug or
>>> register_mem_sect_under_node_early?
>>> Would not that be clean and clear enough?
>>
>> I don't particularly like passing around function pointers where it can
>> be avoided (e.g., here exporting 3 functions now instead 1). Makes the
>> interface harder to get IMHO. But I don't really care about that
>> interface, easy to change later on.
>>
>
> This would lead to the following.
>
> Do everyone agree?
>
> diff --git a/drivers/base/node.c b/drivers/base/node.c
> index 508b80f6329b..444808a7c9b6 100644
> --- a/drivers/base/node.c
> +++ b/drivers/base/node.c
> @@ -761,9 +761,32 @@ static int __ref get_nid_for_pfn(unsigned long pfn)
> return pfn_to_nid(pfn);
> }
>
> +static int do_register_memory_block_under_node(int nid,
> + struct memory_block *mem_blk)
> +{
> + int ret;
> +
> + /*
> + * If this memory block spans multiple nodes, we only indicate
> + * the last processed node.
> + */
> + mem_blk->nid = nid;
> +
> + ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
> + &mem_blk->dev.kobj,
> + kobject_name(&mem_blk->dev.kobj));
> + if (ret)
> + return ret;
> +
> + return sysfs_create_link_nowarn(&mem_blk->dev.kobj,
> + &node_devices[nid]->dev.kobj,
> + kobject_name(&node_devices[nid]->dev.kobj));
> +
> +}
> +
> /* register memory section under specified node if it spans that node */
> -static int register_mem_sect_under_node(struct memory_block *mem_blk,
> - void *arg)
> +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);
> @@ -785,38 +808,35 @@ static int register_mem_sect_under_node(struct
> memory_block *mem_blk,
> }
>
> /*
> - * We need to check if page belongs to nid only for the boot
> - * case, during hotplug we know that all pages in the memory
> - * block belong to the same node.
> + * We need to check if page belongs to nid only at the boot
> + * case because node's ranges can be interleaved.
> */
> - if (system_state == SYSTEM_BOOTING) {
> - page_nid = get_nid_for_pfn(pfn);
> - if (page_nid < 0)
> - continue;
> - if (page_nid != nid)
> - continue;
> - }
> -
> - /*
> - * If this memory block spans multiple nodes, we only indicate
> - * the last processed node.
> - */
> - mem_blk->nid = nid;
> + page_nid = get_nid_for_pfn(pfn);
> + if (page_nid < 0)
> + continue;
> + if (page_nid != nid)
> + continue;
>
> - ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj,
> - &mem_blk->dev.kobj,
> - kobject_name(&mem_blk->dev.kobj));
> + ret = do_register_memory_block_under_node(nid, mem_blk);
> if (ret)
> return ret;
You have to do an unconditional
return ret;
here AFAIKS. For me this looks much better.
Reviewed-by: David Hildenbrand <[email protected]>
--
Thanks,
David / dhildenb