implement "memmap on memory" feature on s390
Subsystems (6)
v5 cover implement "memmap on memory" feature on s390
Hi All,
The patch series implements "memmap on memory" feature on s390.
Patch 1 introduces MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE memory notifiers to prepare the transition of memory to and from a physically accessible state. New mhp_flag MHP_OFFLINE_INACCESSIBLE is introduced to ensure altmap cannot be written when addidng memory - before it is set online. This enhancement is crucial for implementing the "memmap on memory" feature for s390 in a subsequent patch.
Patches 2 allocates vmemmap pages from self-contained memory range for s390. It allocates memory map (struct pages array) from the hotplugged memory range, rather than using system memory by passing altmap to vmemmap functions.
Patch 3 removes unhandled memory notifier types on s390.
Patch 4 implements MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE memory notifiers on s390. MEM_PREPARE_ONLINE memory notifier makes memory block physical accessible via sclp assign command. The notifier ensures self-contained memory maps are accessible and hence enabling the "memmap on memory" on s390. MEM_FINISH_OFFLINE memory notifier shifts the memory block to an inaccessible state via sclp unassign command
Patch 5 finally enables MHP_MEMMAP_ON_MEMORY on s390
Note:
These patches are rebased on top of three fixes:
mm: use vmem_altmap code without CONFIG_ZONE_DEVICE mm/memory_hotplug: fix error handling in add_memory_resource() mm/memory_hotplug: add missing mem_hotplug_lock
v5:
* Added reviewed-by
* Removed variables altmap_start, altmap_size in sclp_cmd.c
* Used PFN_PHYS macro.
Thanks for the valualble feedback.
v4:
* Introduced two new fields, altmap_start_pfn and altmap_nr_pages, in the memory_notify structure and document it that it is used only in MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE callbacks.
* Incorporated the newly added fields into s390's MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE notifier callbacks.
* Prevent access to memblock->altmap->free in the s390 notifier callback.
* page_init_poison() could be performed similar to when adding new memory in sparse_add_section(). Perform it without cond_resched().
v3:
* added comments to MHP_OFFLINE_ACCESSIBLE as suggested by David.
* Squashed three commits related to new memory notifier.
v2:
* Fixes are integrated and hence removed from this patch series Suggestions from David:
* Add new flag MHP_OFFLINE_INACCESSIBLE to avoid accessing memory during memory hotplug addition phase.
* Avoid page_init_poison() on memmap during mhp addition phase, when MHP_OFFLINE_INACCESSIBLE mhp_flag is passed in add_memory().
* Do not skip add_pages() in arch_add_memory(). Similarly, remove similar hacks in arch_remove_memory().
* Use MHP_PREPARE_ONLINE/MHP_FINISH_OFFLINE naming convention for new memory notifiers.
* Rearrange removal of unused s390 memory notifier.
* Necessary commit messages changes.
Thank you
Sumanth Korikkar (5):
mm/memory_hotplug: introduce MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE
notifiers
s390/mm: allocate vmemmap pages from self-contained memory range s390/sclp: remove unhandled memory notifier type s390/mm: implement MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE notifiers
s390: enable MHP_MEMMAP_ON_MEMORY
arch/s390/Kconfig | 1 + arch/s390/mm/init.c | 3 -- arch/s390/mm/vmem.c | 62 +++++++++++++++++++--------------- drivers/base/memory.c | 23 ++++++++++++- drivers/s390/char/sclp_cmd.c | 44 +++++++++++++++++++----- include/linux/memory.h | 9 +++++ include/linux/memory_hotplug.h | 18 +++++++++- include/linux/memremap.h | 1 + mm/memory_hotplug.c | 13 ++++++- mm/sparse.c | 3 +- 10 files changed, 134 insertions(+), 43 deletions(-)
Signature · 1 line
2.40.1
PATCH v5 1/5 mm/memory_hotplug: introduce MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE notifiers
Introduce MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE memory notifiers to prepare the transition of memory to and from a physically accessible state. This enhancement is crucial for implementing the "memmap on memory" feature for s390 in a subsequent patch.
Platforms such as x86 can support physical memory hotplug via ACPI. When there is physical memory hotplug, ACPI event leads to the memory addition with the following callchain:acpi_memory_device_add()
-> acpi_memory_enable_device()
-> __add_memory()
After this, the hotplugged memory is physically accessible, and altmap support prepared, before the "memmap on memory" initialization in memory_block_online() is called.
On s390, memory hotplug works in a different way. The available hotplug memory has to be defined upfront in the hypervisor, but it is made physically accessible only when the user sets it online via sysfs, currently in the MEM_GOING_ONLINE notifier. This is too late and "memmap on memory" initialization is performed before calling MEM_GOING_ONLINE notifier.
During the memory hotplug addition phase, altmap support is prepared and during the memory onlining phase s390 requires memory to be physically accessible and then subsequently initiate the "memmap on memory" initialization process.
The memory provider will handle new MEM_PREPARE_ONLINE / MEM_FINISH_OFFLINE notifications and make the memory accessible.
The mhp_flag MHP_OFFLINE_INACCESSIBLE is introduced and is relevant when used along with MHP_MEMMAP_ON_MEMORY, because the altmap cannot be written (e.g., poisoned) when adding memory -- before it is set online. This allows for adding memory with an altmap that is not currently made available by a hypervisor. When onlining that memory, the hypervisor can be instructed to make that memory accessible via the new notifiers and the onlining phase will not require any memory allocations, which is helpful in low-memory situations.
All architectures ignore unknown memory notifiers. Therefore, the introduction of these new notifiers does not result in any functional modifications across architectures.
--- drivers/base/memory.c | 23 ++++++++++++++++++++++- include/linux/memory.h | 9 +++++++++ include/linux/memory_hotplug.h | 18 +++++++++++++++++- include/linux/memremap.h | 1 + mm/memory_hotplug.c | 13 ++++++++++++- mm/sparse.c | 3 ++- 6 files changed, 63 insertions(+), 4 deletions(-)
drivers/base/memory.c +22 -1
@@ -188,6 +188,7 @@ 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 nr_vmemmap_pages =0 ; +struct memory_notify arg;struct zone *zone;int ret;
@@ -207,9 +208,19 @@ static int memory_block_online(struct memory_block *mem)if (mem->altmap) nr_vmemmap_pages = mem->altmap->free; + arg.altmap_start_pfn = start_pfn; + arg.altmap_nr_pages = nr_vmemmap_pages; + arg.start_pfn = start_pfn + nr_vmemmap_pages; + arg.nr_pages = nr_pages - nr_vmemmap_pages;mem_hotplug_begin (); + ret =memory_notify (MEM_PREPARE_ONLINE , &arg); + ret =notifier_to_errno (ret); +if (ret) + goto out_notifier; +if (nr_vmemmap_pages) { - ret =mhp_init_memmap_on_memory (start_pfn, nr_vmemmap_pages, zone); + ret =mhp_init_memmap_on_memory (start_pfn, nr_vmemmap_pages, + zone, mem->altmap->inaccessible);if (ret) goto out; }
@@ -231,7 +242,11 @@ static int memory_block_online(struct memory_block *mem) nr_vmemmap_pages); mem->zone = zone; +mem_hotplug_done (); +return ret; out: +memory_notify (MEM_FINISH_OFFLINE , &arg); +out_notifier:mem_hotplug_done ();return ret; }
@@ -244,6 +259,7 @@ 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 nr_vmemmap_pages =0 ; +struct memory_notify arg;int ret;if (!mem->zone)
@@ -275,6 +291,11 @@ static int memory_block_offline(struct memory_block *mem)mhp_deinit_memmap_on_memory (start_pfn, nr_vmemmap_pages); mem->zone =NULL ; + arg.altmap_start_pfn = start_pfn; + arg.altmap_nr_pages = nr_vmemmap_pages; + arg.start_pfn = start_pfn + nr_vmemmap_pages; + arg.nr_pages = nr_pages - nr_vmemmap_pages; +memory_notify (MEM_FINISH_OFFLINE , &arg); out:mem_hotplug_done ();return ret;
include/linux/memory.h +9 -0
@@ -96,8 +96,17 @@ int set_memory_block_size_order(unsigned int order);#define MEM_GOING_ONLINE (1<<3)#define MEM_CANCEL_ONLINE (1<<4)#define MEM_CANCEL_OFFLINE (1<<5) +#define MEM_PREPARE_ONLINE (1<<6) +#define MEM_FINISH_OFFLINE (1<<7)struct memory_notify { +/* +* The altmap_start_pfn and altmap_nr_pages fields are designated for +* specifying the altmap range and are exclusively intended for use in +* MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE notifiers. +*/ +unsigned long altmap_start_pfn; +unsigned long altmap_nr_pages;unsigned long start_pfn;unsigned long nr_pages;int status_change_nid_normal;
include/linux/memory_hotplug.h +17 -1
@@ -106,6 +106,22 @@ typedef int __bitwise mhp_t;* implies the node id (nid). */ #define MHP_NID_IS_MGID ((__force mhp_t)BIT(2)) +/* +* The hotplugged memory is completely inaccessible while the memory is +* offline. The memory provider will handle MEM_PREPARE_ONLINE / +* MEM_FINISH_OFFLINE notifications and make the memory accessible. +* +* This flag is only relevant when used along with MHP_MEMMAP_ON_MEMORY, +* because the altmap cannot be written (e.g., poisoned) when adding +* memory -- before it is set online. +* +* This allows for adding memory with an altmap that is not currently +* made available by a hypervisor. When onlining that memory, the +* hypervisor can be instructed to make that memory available, and +* the onlining phase will not require any memory allocations, which is +* helpful in low-memory situations. +*/ +#define MHP_OFFLINE_INACCESSIBLE ((__force mhp_t)BIT(3))/* * Extended parameters for memory hotplug:
@@ -154,7 +170,7 @@ extern void adjust_present_page_count(struct page *page, long nr_pages);/* VM interface that may be used by firmware interface */ extern int mhp_init_memmap_on_memory (unsigned long pfn,unsigned long nr_pages, -struct zone *zone); +struct zone *zone,bool mhp_off_inaccessible);extern void mhp_deinit_memmap_on_memory (unsigned long pfn,unsigned long nr_pages);extern int online_pages (unsigned long pfn,unsigned long nr_pages,struct zone *zone,struct memory_group *group);
include/linux/memremap.h +1 -0
@@ -25,6 +25,7 @@ struct vmem_altmap {unsigned long free;unsigned long align;unsigned long alloc; +bool inaccessible; };/*
mm/memory_hotplug.c +12 -1
@@ -1084,7 +1084,7 @@ void adjust_present_page_count(struct page *page, struct memory_group *group, }int mhp_init_memmap_on_memory (unsigned long pfn,unsigned long nr_pages, -struct zone *zone) +struct zone *zone,bool mhp_off_inaccessible) {unsigned long end_pfn = pfn + nr_pages;int ret, i;
@@ -1093,6 +1093,15 @@ int mhp_init_memmap_on_memory(unsigned long pfn, unsigned long nr_pages,if (ret)return ret; +/* +* Memory block is accessible at this stage and hence poison the struct +* pages now. If the memory block is accessible during memory hotplug +* addition phase, then page poisining is already performed in +* sparse_add_section(). +*/ +if (mhp_off_inaccessible) +page_init_poison (pfn_to_page (pfn),sizeof (struct page ) * nr_pages); +move_pfn_range_to_zone (zone, pfn, nr_pages,NULL ,MIGRATE_UNMOVABLE );for (i =0 ; i < nr_pages; i++)
@@ -1439,6 +1448,8 @@ int __ref add_memory_resource(int nid, struct resource *res, mhp_t mhp_flags)if (mhp_flags &MHP_MEMMAP_ON_MEMORY ) {if (mhp_supports_memmap_on_memory (size)) { mhp_altmap.free =memory_block_memmap_on_memory_pages (); +if (mhp_flags &MHP_OFFLINE_INACCESSIBLE ) + mhp_altmap.inaccessible = true; params.altmap =kmalloc (sizeof (struct vmem_altmap ),GFP_KERNEL );if (!params.altmap) { ret = -ENOMEM ;
mm/sparse.c +2 -1
@@ -907,7 +907,8 @@ int __meminit sparse_add_section(int nid, unsigned long start_pfn,* Poison uninitialized struct pages in order to catch invalid flags * combinations. */ -page_init_poison (memmap,sizeof (struct page ) * nr_pages); +if (!altmap || !altmap->inaccessible) +page_init_poison (memmap,sizeof (struct page ) * nr_pages); ms =__nr_to_section (section_nr);set_section_nid (section_nr, nid);
Signature · 1 line
2.40.1
PATCH v5 2/5 s390/mm: allocate vmemmap pages from self-contained memory range
In reply to Sumanth KorikkarAllocate memory map (struct pages array) from the hotplugged memory range, rather than using system memory. The change addresses the issue where standby memory, when configured to be much larger than online memory, could potentially lead to ipl failure due to memory map allocation from online memory. For example, 16MB of memory map allocation is needed for a memory block size of 1GB and when standby memory is configured much larger than online memory, this could lead to ipl failure.
To address this issue, the solution involves introducing "memmap on memory" using the vmem_altmap structure on s390. Architectures that want to implement it should pass the altmap to the vmemmap_populate() function and its associated callchain. This enhancement is discussed in the commit 4b94ffdc4163 ("x86, mm: introduce vmem_altmap to augment vmemmap_populate()").
Provide "memmap on memory" support for s390 by passing the altmap in vmemmap_populate() and its callchain. The allocation path is described as follows:
* When altmap is NULL in vmemmap_populate(), memory map allocation occurs using the existing vmemmap_alloc_block_buf().
* When altmap is not NULL in vmemmap_populate(), memory map allocation still uses vmemmap_alloc_block_buf(), but this function internally calls altmap_alloc_block_buf().
For deallocation, the process is outlined as follows:
* When altmap is NULL in vmemmap_free(), memory map deallocation happens through free_pages().
* When altmap is not NULL in vmemmap_free(), memory map deallocation occurs via vmem_altmap_free().
While memory map allocation is primarily handled through the self-contained memory map range, there might still be a small amount of system memory allocation required for vmemmap pagetables. To mitigate this impact, this feature will be limited to machines with EDAT1 support.
--- arch/s390/mm/init.c | 3 --- arch/s390/mm/vmem.c | 62 +++++++++++++++++++++++++-------------------- 2 files changed, 35 insertions(+), 30 deletions(-)
arch/s390/mm/init.c +0 -3
@@ -281,9 +281,6 @@ int arch_add_memory(int nid, u64 start, u64 size,unsigned long size_pages =PFN_DOWN (size);int rc; -if (WARN_ON_ONCE (params->altmap)) -return -EINVAL ; -if (WARN_ON_ONCE (params->pgprot.pgprot !=PAGE_KERNEL .pgprot))return -EINVAL ;
arch/s390/mm/vmem.c +35 -27
194 lines not on this page · raw message
Signature · 1 line
2.40.1
PATCH v5 3/5 s390/sclp: remove unhandled memory notifier type
In reply to Sumanth KorikkarRemove memory notifier types which are unhandled by s390. Unhandled memory notifier types are covered by default case.
--- drivers/s390/char/sclp_cmd.c | 3 --- 1 file changed, 3 deletions(-)
drivers/s390/char/sclp_cmd.c +0 -3
@@ -340,9 +340,6 @@ static int sclp_mem_notifier(struct notifier_block *nb,if (contains_standby_increment (start, start + size)) rc = -EPERM ;break ; -case MEM_ONLINE : -case MEM_CANCEL_OFFLINE : -break ;case MEM_GOING_ONLINE : rc =sclp_mem_change_state (start, size,1 );break ;
Signature · 1 line
2.40.1
PATCH v5 4/5 s390/mm: implement MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE notifiers
In reply to Sumanth KorikkarMEM_PREPARE_ONLINE memory notifier makes memory block physical accessible via sclp assign command. The notifier ensures self-contained memory maps are accessible and hence enabling the "memmap on memory" on s390.
MEM_FINISH_OFFLINE memory notifier shifts the memory block to an inaccessible state via sclp unassign command.
Implementation considerations:
* When MHP_MEMMAP_ON_MEMORY is disabled, the system retains the old behavior. This means the memory map is allocated from default memory.
* If MACHINE_HAS_EDAT1 is unavailable, MHP_MEMMAP_ON_MEMORY is automatically disabled. This ensures that vmemmap pagetables do not consume additional memory from the default memory allocator.
* The MEM_GOING_ONLINE notifier has been modified to perform no operation, as MEM_PREPARE_ONLINE already executes the sclp assign command.
* The MEM_CANCEL_ONLINE/MEM_OFFLINE notifier now performs no operation, as MEM_FINISH_OFFLINE already executes the sclp unassign command.
--- drivers/s390/char/sclp_cmd.c | 41 ++++++++++++++++++++++++++++++------ 1 file changed, 35 insertions(+), 6 deletions(-)
drivers/s390/char/sclp_cmd.c +35 -6
@@ -18,6 +18,7 @@#include <linux/mm.h> #include <linux/mmzone.h> #include <linux/memory.h> +#include <linux/memory_hotplug.h> #include <linux/module.h> #include <asm/ctlreg.h> #include <asm/chpid.h>
@@ -26,6 +27,7 @@#include <asm/sclp.h> #include <asm/numa.h> #include <asm/facility.h> +#include <asm/page-states.h> #include "sclp.h"
@@ -340,13 +342,38 @@ static int sclp_mem_notifier(struct notifier_block *nb,if (contains_standby_increment (start, start + size)) rc = -EPERM ;break ; -case MEM_GOING_ONLINE : +case MEM_PREPARE_ONLINE : +/* +* Access the altmap_start_pfn and altmap_nr_pages fields +* within the struct memory_notify specifically when dealing +* with only MEM_PREPARE_ONLINE/MEM_FINISH_OFFLINE notifiers. +* +* When altmap is in use, take the specified memory range +* online, which includes the altmap. +*/ +if (arg->altmap_nr_pages) { + start =PFN_PHYS (arg->altmap_start_pfn); + size +=PFN_PHYS (arg->altmap_nr_pages); + } rc =sclp_mem_change_state (start, size,1 ); +if (rc || !arg->altmap_nr_pages) +break ; +/* +* Set CMMA state to nodat here, since the struct page memory +* at the beginning of the memory block will not go through the +* buddy allocator later. +*/ +__arch_set_page_nodat ((void *)__va (start), arg->altmap_nr_pages);break ; -case MEM_CANCEL_ONLINE : -sclp_mem_change_state (start, size,0 ); -break ; -case MEM_OFFLINE : +case MEM_FINISH_OFFLINE : +/* +* When altmap is in use, take the specified memory range +* offline, which includes the altmap. +*/ +if (arg->altmap_nr_pages) { + start =PFN_PHYS (arg->altmap_start_pfn); + size +=PFN_PHYS (arg->altmap_nr_pages); + }sclp_mem_change_state (start, size,0 );break ;default :
@@ -397,7 +424,9 @@ static void __init add_memory_merged(u16 rn)if (!size) goto skip_add;for (addr = start; addr < start + size; addr += block_size) -add_memory (0 , addr, block_size,MHP_NONE ); +add_memory (0 , addr, block_size, +MACHINE_HAS_EDAT1 ? +MHP_MEMMAP_ON_MEMORY |MHP_OFFLINE_INACCESSIBLE :MHP_NONE ); skip_add: first_rn = rn; num =1 ;
Signature · 1 line
2.40.1
PATCH v5 5/5 s390: enable MHP_MEMMAP_ON_MEMORY
In reply to Sumanth KorikkarEnable MHP_MEMMAP_ON_MEMORY to support "memmap on memory". memory_hotplug.memmap_on_memory=true kernel parameter should be set in kernel boot option to enable the feature.
--- arch/s390/Kconfig | 1 + 1 file changed, 1 insertion(+)
arch/s390/Kconfig +1 -0
@@ -113,6 +113,7 @@ config S390 select ARCH_INLINE_WRITE_UNLOCK_BH select ARCH_INLINE_WRITE_UNLOCK_IRQ select ARCH_INLINE_WRITE_UNLOCK_IRQRESTORE + select ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE select ARCH_STACKWALK select ARCH_SUPPORTS_ATOMIC_RMW select ARCH_SUPPORTS_DEBUG_PAGEALLOC
Signature · 1 line
2.40.1
Andrew Morton
maintainerQuoted 1 line · On Tue, 28 Nov 2023 16:52:22 +0100 Sumanth Korikkar <[email protected]>…
The patch series implements "memmap on memory" feature on s390.
The cover letter doesn't acutally have a description of what "memmap on memory" *is*. A nice overview to help readers understand what they're about to look at. A description of what value this feature brings to our users. Use-cases. That sort of thing.
I guess the [1/N] changelog covers it, but it's hard to tell. It isn't exactly broad-sweep overview.
Probably something short would suffice. There are plenty of examples on the mailing list, please take a look and send us something?
Alexander Gordeev
maintainerOn Tue, Nov 28, 2023 at 03:56:34PM -0800, Andrew Morton wrote:
Hi Andrew,
Quoted 10 lines
The cover letter doesn't acutally have a description of what "memmap on memory" *is*. A nice overview to help readers understand what they're about to look at. A description of what value this feature brings to our users. Use-cases. That sort of thing.
I guess the [1/N] changelog covers it, but it's hard to tell. It isn't exactly broad-sweep overview.
Probably something short would suffice. There are plenty of examples on the mailing list, please take a look and send us something?
Sumanth is on vacation, I will try to answer.
This series brings "memmap on memory" support to s390 platform. That is - allocate 'struct pages' array describing a memory block being onlined not from available system memory, but from the memory block itself.
This improves the current situation on s390, where struct pages for all memory that potentially can be added to the system, is preallocated, so that memory online will always succeed but at the cost of significant memory consumption.
Unlike other architectures, the challenge s390 faces is the memory block being onlined is not accessible at the time of hotplug event. To make it physically accessible two new MEM_PREPARE_ONLINE and MEM_FINISH_OFFLINE memory notifiers are added. That allows to succeed "memmap on memory" initialization.
For the existing architectures the two new notifiers are unknown and they are not affected in any way.
Thanks!