From e930d999715073a70d306fb59a394ea8b84d0b45 Mon Sep 17 00:00:00 2001 From: Michal Hocko Date: Tue, 22 Mar 2022 14:46:51 -0700 Subject: mm, memory_hotplug: make arch_alloc_nodedata independent on CONFIG_MEMORY_HOTPLUG Patch series "mm, memory_hotplug: handle unitialized numa node gracefully". The core of the fix is patch 2 which also links existing bug reports. The high level goal is to have all possible numa nodes have their pgdat allocated and initialized so for_each_possible_node(nid) NODE_DATA(nid) will never return garbage. This has proven to be problem in several places when an offline numa node is used for an allocation just to realize that node_data and therefore allocation fallback zonelists are not initialized and such an allocation request blows up. There were attempts to address that by checking node_online in several places including the page allocator. This patchset approaches the problem from a different perspective and instead of special casing, which just adds a runtime overhead, it allocates pglist_data for each possible node. This can add some memory overhead for platforms with high number of possible nodes if they do not contain any memory. This should be a rather rare configuration though. How to test this? David has provided and excellent howto: http://lkml.kernel.org/r/6e5ebc19-890c-b6dd-1924-9f25c441010d@redhat.com Patches 1 and 3-6 are mostly cleanups. The patchset has been reviewed by Rafael (thanks!) and the core fix tested by Rafael and Alexey (thanks to both). David has tested as per instructions above and hasn't found any fallouts in the memory hotplug scenarios. This patch (of 6): This is a preparatory patch and it doesn't introduce any functional change. It merely pulls out arch_alloc_nodedata (and co) outside of CONFIG_MEMORY_HOTPLUG because the following patch will need to call this from the generic MM code. Link: https://lkml.kernel.org/r/20220127085305.20890-1-mhocko@kernel.org Link: https://lkml.kernel.org/r/20220127085305.20890-2-mhocko@kernel.org Signed-off-by: Michal Hocko Acked-by: Rafael Aquini Acked-by: David Hildenbrand Acked-by: Mike Rapoport Reviewed-by: Oscar Salvador Reviewed-by: Wei Yang Cc: Alexey Makhalov Cc: Christoph Lameter Cc: Dennis Zhou Cc: Eric Dumazet Cc: Nico Pache Cc: Tejun Heo Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memory_hotplug.h | 119 ++++++++++++++++++++--------------------- 1 file changed, 59 insertions(+), 60 deletions(-) (limited to 'include/linux/memory_hotplug.h') diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h index be48e003a518..4355983b364d 100644 --- a/include/linux/memory_hotplug.h +++ b/include/linux/memory_hotplug.h @@ -16,6 +16,65 @@ struct memory_group; struct resource; struct vmem_altmap; +#ifdef CONFIG_HAVE_ARCH_NODEDATA_EXTENSION +/* + * For supporting node-hotadd, we have to allocate a new pgdat. + * + * If an arch has generic style NODE_DATA(), + * node_data[nid] = kzalloc() works well. But it depends on the architecture. + * + * In general, generic_alloc_nodedata() is used. + * Now, arch_free_nodedata() is just defined for error path of node_hot_add. + * + */ +extern pg_data_t *arch_alloc_nodedata(int nid); +extern void arch_free_nodedata(pg_data_t *pgdat); +extern void arch_refresh_nodedata(int nid, pg_data_t *pgdat); + +#else /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */ + +#define arch_alloc_nodedata(nid) generic_alloc_nodedata(nid) +#define arch_free_nodedata(pgdat) generic_free_nodedata(pgdat) + +#ifdef CONFIG_NUMA +/* + * XXX: node aware allocation can't work well to get new node's memory at this time. + * Because, pgdat for the new node is not allocated/initialized yet itself. + * To use new node's memory, more consideration will be necessary. + */ +#define generic_alloc_nodedata(nid) \ +({ \ + kzalloc(sizeof(pg_data_t), GFP_KERNEL); \ +}) +/* + * This definition is just for error path in node hotadd. + * For node hotremove, we have to replace this. + */ +#define generic_free_nodedata(pgdat) kfree(pgdat) + +extern pg_data_t *node_data[]; +static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat) +{ + node_data[nid] = pgdat; +} + +#else /* !CONFIG_NUMA */ + +/* never called */ +static inline pg_data_t *generic_alloc_nodedata(int nid) +{ + BUG(); + return NULL; +} +static inline void generic_free_nodedata(pg_data_t *pgdat) +{ +} +static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat) +{ +} +#endif /* CONFIG_NUMA */ +#endif /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */ + #ifdef CONFIG_MEMORY_HOTPLUG struct page *pfn_to_online_page(unsigned long pfn); @@ -154,66 +213,6 @@ int add_pages(int nid, unsigned long start_pfn, unsigned long nr_pages, struct mhp_params *params); #endif /* ARCH_HAS_ADD_PAGES */ -#ifdef CONFIG_HAVE_ARCH_NODEDATA_EXTENSION -/* - * For supporting node-hotadd, we have to allocate a new pgdat. - * - * If an arch has generic style NODE_DATA(), - * node_data[nid] = kzalloc() works well. But it depends on the architecture. - * - * In general, generic_alloc_nodedata() is used. - * Now, arch_free_nodedata() is just defined for error path of node_hot_add. - * - */ -extern pg_data_t *arch_alloc_nodedata(int nid); -extern void arch_free_nodedata(pg_data_t *pgdat); -extern void arch_refresh_nodedata(int nid, pg_data_t *pgdat); - -#else /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */ - -#define arch_alloc_nodedata(nid) generic_alloc_nodedata(nid) -#define arch_free_nodedata(pgdat) generic_free_nodedata(pgdat) - -#ifdef CONFIG_NUMA -/* - * If ARCH_HAS_NODEDATA_EXTENSION=n, this func is used to allocate pgdat. - * XXX: kmalloc_node() can't work well to get new node's memory at this time. - * Because, pgdat for the new node is not allocated/initialized yet itself. - * To use new node's memory, more consideration will be necessary. - */ -#define generic_alloc_nodedata(nid) \ -({ \ - kzalloc(sizeof(pg_data_t), GFP_KERNEL); \ -}) -/* - * This definition is just for error path in node hotadd. - * For node hotremove, we have to replace this. - */ -#define generic_free_nodedata(pgdat) kfree(pgdat) - -extern pg_data_t *node_data[]; -static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat) -{ - node_data[nid] = pgdat; -} - -#else /* !CONFIG_NUMA */ - -/* never called */ -static inline pg_data_t *generic_alloc_nodedata(int nid) -{ - BUG(); - return NULL; -} -static inline void generic_free_nodedata(pg_data_t *pgdat) -{ -} -static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat) -{ -} -#endif /* CONFIG_NUMA */ -#endif /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */ - void get_online_mems(void); void put_online_mems(void); -- cgit v1.2.3-71-gd317 From 09f49dca570a917a8c6bccd7e8c61f5141534e3a Mon Sep 17 00:00:00 2001 From: Michal Hocko Date: Tue, 22 Mar 2022 14:46:54 -0700 Subject: mm: handle uninitialized numa nodes gracefully We have had several reports [1][2][3] that page allocator blows up when an allocation from a possible node is requested. The underlying reason is that NODE_DATA for the specific node is not allocated. NUMA specific initialization is arch specific and it can vary a lot. E.g. x86 tries to initialize all nodes that have some cpu affinity (see init_cpu_to_node) but this can be insufficient because the node might be cpuless for example. One way to address this problem would be to check for !node_online nodes when trying to get a zonelist and silently fall back to another node. That is unfortunately adding a branch into allocator hot path and it doesn't handle any other potential NODE_DATA users. This patch takes a different approach (following a lead of [3]) and it pre allocates pgdat for all possible nodes in an arch indipendent code - free_area_init. All uninitialized nodes are treated as memoryless nodes. node_state of the node is not changed because that would lead to other side effects - e.g. sysfs representation of such a node and from past discussions [4] it is known that some tools might have problems digesting that. Newly allocated pgdat only gets a minimal initialization and the rest of the work is expected to be done by the memory hotplug - hotadd_new_pgdat (renamed to hotadd_init_pgdat). generic_alloc_nodedata is changed to use the memblock allocator because neither page nor slab allocators are available at the stage when all pgdats are allocated. Hotplug doesn't allocate pgdat anymore so we can use the early boot allocator. The only arch specific implementation is ia64 and that is changed to use the early allocator as well. [1] http://lkml.kernel.org/r/20211101201312.11589-1-amakhalov@vmware.com [2] http://lkml.kernel.org/r/20211207224013.880775-1-npache@redhat.com [3] http://lkml.kernel.org/r/20190114082416.30939-1-mhocko@kernel.org [4] http://lkml.kernel.org/r/20200428093836.27190-1-srikar@linux.vnet.ibm.com [akpm@linux-foundation.org: replace comment, per Mike] Link: https://lkml.kernel.org/r/Yfe7RBeLCijnWBON@dhcp22.suse.cz Reported-by: Alexey Makhalov Tested-by: Alexey Makhalov Reported-by: Nico Pache Acked-by: Rafael Aquini Tested-by: Rafael Aquini Acked-by: David Hildenbrand Reviewed-by: Oscar Salvador Acked-by: Mike Rapoport Signed-off-by: Michal Hocko Cc: Christoph Lameter Cc: Dennis Zhou Cc: Eric Dumazet Cc: Tejun Heo Cc: Wei Yang Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/ia64/mm/discontig.c | 4 ++-- include/linux/memory_hotplug.h | 2 +- mm/internal.h | 2 ++ mm/memory_hotplug.c | 21 +++++++++------------ mm/page_alloc.c | 40 ++++++++++++++++++++++++++++++++++++---- 5 files changed, 50 insertions(+), 19 deletions(-) (limited to 'include/linux/memory_hotplug.h') diff --git a/arch/ia64/mm/discontig.c b/arch/ia64/mm/discontig.c index 8dc8a554f774..dd0cf4834eaa 100644 --- a/arch/ia64/mm/discontig.c +++ b/arch/ia64/mm/discontig.c @@ -608,11 +608,11 @@ void __init paging_init(void) zero_page_memmap_ptr = virt_to_page(ia64_imva(empty_zero_page)); } -pg_data_t *arch_alloc_nodedata(int nid) +pg_data_t * __init arch_alloc_nodedata(int nid) { unsigned long size = compute_pernodesize(nid); - return kzalloc(size, GFP_KERNEL); + return memblock_alloc(size, SMP_CACHE_BYTES); } void arch_free_nodedata(pg_data_t *pgdat) diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h index 4355983b364d..cdd66bfdf855 100644 --- a/include/linux/memory_hotplug.h +++ b/include/linux/memory_hotplug.h @@ -44,7 +44,7 @@ extern void arch_refresh_nodedata(int nid, pg_data_t *pgdat); */ #define generic_alloc_nodedata(nid) \ ({ \ - kzalloc(sizeof(pg_data_t), GFP_KERNEL); \ + memblock_alloc(sizeof(*pgdat), SMP_CACHE_BYTES); \ }) /* * This definition is just for error path in node hotadd. diff --git a/mm/internal.h b/mm/internal.h index 9c298afb9688..f1554a4e249e 100644 --- a/mm/internal.h +++ b/mm/internal.h @@ -707,4 +707,6 @@ void vunmap_range_noflush(unsigned long start, unsigned long end); int numa_migrate_prep(struct page *page, struct vm_area_struct *vma, unsigned long addr, int page_nid, int *flags); +DECLARE_PER_CPU(struct per_cpu_nodestat, boot_nodestats); + #endif /* __MM_INTERNAL_H */ diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c index 0139b77c51d5..11f39d0e76ec 100644 --- a/mm/memory_hotplug.c +++ b/mm/memory_hotplug.c @@ -1162,19 +1162,21 @@ static void reset_node_present_pages(pg_data_t *pgdat) } /* we are OK calling __meminit stuff here - we have CONFIG_MEMORY_HOTPLUG */ -static pg_data_t __ref *hotadd_new_pgdat(int nid) +static pg_data_t __ref *hotadd_init_pgdat(int nid) { struct pglist_data *pgdat; pgdat = NODE_DATA(nid); - if (!pgdat) { - pgdat = arch_alloc_nodedata(nid); - if (!pgdat) - return NULL; + /* + * NODE_DATA is preallocated (free_area_init) but its internal + * state is not allocated completely. Add missing pieces. + * Completely offline nodes stay around and they just need + * reintialization. + */ + if (pgdat->per_cpu_nodestats == &boot_nodestats) { pgdat->per_cpu_nodestats = alloc_percpu(struct per_cpu_nodestat); - arch_refresh_nodedata(nid, pgdat); } else { int cpu; /* @@ -1193,8 +1195,6 @@ static pg_data_t __ref *hotadd_new_pgdat(int nid) } } - /* we can use NODE_DATA(nid) from here */ - pgdat->node_id = nid; pgdat->node_start_pfn = 0; /* init node's zones as empty zones, we don't have any present pages.*/ @@ -1246,7 +1246,7 @@ static int __try_online_node(int nid, bool set_node_online) if (node_online(nid)) return 0; - pgdat = hotadd_new_pgdat(nid); + pgdat = hotadd_init_pgdat(nid); if (!pgdat) { pr_err("Cannot online node %d due to NULL pgdat\n", nid); ret = -ENOMEM; @@ -1445,9 +1445,6 @@ int __ref add_memory_resource(int nid, struct resource *res, mhp_t mhp_flags) return ret; error: - /* rollback pgdat allocation and others */ - if (new_node) - rollback_node_hotadd(nid); if (IS_ENABLED(CONFIG_ARCH_KEEP_MEMBLOCK)) memblock_remove(start, size); error_mem_hotplug_end: diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 8b18a077c409..4f141a4e5b64 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -6341,7 +6341,7 @@ static void per_cpu_pages_init(struct per_cpu_pages *pcp, struct per_cpu_zonesta #define BOOT_PAGESET_BATCH 1 static DEFINE_PER_CPU(struct per_cpu_pages, boot_pageset); static DEFINE_PER_CPU(struct per_cpu_zonestat, boot_zonestats); -static DEFINE_PER_CPU(struct per_cpu_nodestat, boot_nodestats); +DEFINE_PER_CPU(struct per_cpu_nodestat, boot_nodestats); static void __build_all_zonelists(void *data) { @@ -6363,7 +6363,11 @@ static void __build_all_zonelists(void *data) if (self && !node_online(self->node_id)) { build_zonelists(self); } else { - for_each_online_node(nid) { + /* + * All possible nodes have pgdat preallocated + * in free_area_init + */ + for_each_node(nid) { pg_data_t *pgdat = NODE_DATA(nid); build_zonelists(pgdat); @@ -8063,8 +8067,36 @@ void __init free_area_init(unsigned long *max_zone_pfn) /* Initialise every node */ mminit_verify_pageflags_layout(); setup_nr_node_ids(); - for_each_online_node(nid) { - pg_data_t *pgdat = NODE_DATA(nid); + for_each_node(nid) { + pg_data_t *pgdat; + + if (!node_online(nid)) { + pr_info("Initializing node %d as memoryless\n", nid); + + /* Allocator not initialized yet */ + pgdat = arch_alloc_nodedata(nid); + if (!pgdat) { + pr_err("Cannot allocate %zuB for node %d.\n", + sizeof(*pgdat), nid); + continue; + } + arch_refresh_nodedata(nid, pgdat); + free_area_init_memoryless_node(nid); + + /* + * We do not want to confuse userspace by sysfs + * files/directories for node without any memory + * attached to it, so this node is not marked as + * N_MEMORY and not marked online so that no sysfs + * hierarchy will be created via register_one_node for + * it. The pgdat will get fully initialized by + * hotadd_init_pgdat() when memory is hotplugged into + * this node. + */ + continue; + } + + pgdat = NODE_DATA(nid); free_area_init_node(nid); /* Any memory on that node */ -- cgit v1.2.3-71-gd317 From 390511e1476eb1cc41d420a7661b33f4d8584c3f Mon Sep 17 00:00:00 2001 From: Michal Hocko Date: Tue, 22 Mar 2022 14:46:57 -0700 Subject: mm, memory_hotplug: drop arch_free_nodedata Prior to "mm: handle uninitialized numa nodes gracefully" memory hotplug used to allocate pgdat when memory has been added to a node (hotadd_init_pgdat) arch_free_nodedata has been only used in the failure path because once the pgdat is exported (to be visible by NODA_DATA(nid)) it cannot really be freed because there is no synchronization available for that. pgdat is allocated for each possible nodes now so the memory hotplug doesn't need to do the ever use arch_free_nodedata so drop it. This patch doesn't introduce any functional change. Link: https://lkml.kernel.org/r/20220127085305.20890-4-mhocko@kernel.org Signed-off-by: Michal Hocko Acked-by: Rafael Aquini Acked-by: David Hildenbrand Acked-by: Mike Rapoport Reviewed-by: Oscar Salvador Cc: Alexey Makhalov Cc: Christoph Lameter Cc: Dennis Zhou Cc: Eric Dumazet Cc: Nico Pache Cc: Tejun Heo Cc: Wei Yang Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/ia64/mm/discontig.c | 5 ----- include/linux/memory_hotplug.h | 3 --- mm/memory_hotplug.c | 10 ---------- 3 files changed, 18 deletions(-) (limited to 'include/linux/memory_hotplug.h') diff --git a/arch/ia64/mm/discontig.c b/arch/ia64/mm/discontig.c index dd0cf4834eaa..73d0db36edb6 100644 --- a/arch/ia64/mm/discontig.c +++ b/arch/ia64/mm/discontig.c @@ -615,11 +615,6 @@ pg_data_t * __init arch_alloc_nodedata(int nid) return memblock_alloc(size, SMP_CACHE_BYTES); } -void arch_free_nodedata(pg_data_t *pgdat) -{ - kfree(pgdat); -} - void arch_refresh_nodedata(int update_node, pg_data_t *update_pgdat) { pgdat_list[update_node] = update_pgdat; diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h index cdd66bfdf855..60f09d3ebb3d 100644 --- a/include/linux/memory_hotplug.h +++ b/include/linux/memory_hotplug.h @@ -24,17 +24,14 @@ struct vmem_altmap; * node_data[nid] = kzalloc() works well. But it depends on the architecture. * * In general, generic_alloc_nodedata() is used. - * Now, arch_free_nodedata() is just defined for error path of node_hot_add. * */ extern pg_data_t *arch_alloc_nodedata(int nid); -extern void arch_free_nodedata(pg_data_t *pgdat); extern void arch_refresh_nodedata(int nid, pg_data_t *pgdat); #else /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */ #define arch_alloc_nodedata(nid) generic_alloc_nodedata(nid) -#define arch_free_nodedata(pgdat) generic_free_nodedata(pgdat) #ifdef CONFIG_NUMA /* diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c index 11f39d0e76ec..55c3e5309088 100644 --- a/mm/memory_hotplug.c +++ b/mm/memory_hotplug.c @@ -1217,16 +1217,6 @@ static pg_data_t __ref *hotadd_init_pgdat(int nid) return pgdat; } -static void rollback_node_hotadd(int nid) -{ - pg_data_t *pgdat = NODE_DATA(nid); - - arch_refresh_nodedata(nid, NULL); - free_percpu(pgdat->per_cpu_nodestats); - arch_free_nodedata(pgdat); -} - - /* * __try_online_node - online a node if offlined * @nid: the node ID -- cgit v1.2.3-71-gd317 From 70b5b46a754245d383811b4d2f2c76c34bb7e145 Mon Sep 17 00:00:00 2001 From: Michal Hocko Date: Tue, 22 Mar 2022 14:47:00 -0700 Subject: mm, memory_hotplug: reorganize new pgdat initialization When a !node_online node is brought up it needs a hotplug specific initialization because the node could be either uninitialized yet or it could have been recycled after previous hotremove. hotadd_init_pgdat is responsible for that. Internal pgdat state is initialized at two places currently - hotadd_init_pgdat - free_area_init_core_hotplug There is no real clear cut what should go where but this patch's chosen to move the whole internal state initialization into free_area_init_core_hotplug. hotadd_init_pgdat is still responsible to pull all the parts together - most notably to initialize zonelists because those depend on the overall topology. This patch doesn't introduce any functional change. Link: https://lkml.kernel.org/r/20220127085305.20890-5-mhocko@kernel.org Signed-off-by: Michal Hocko Acked-by: Rafael Aquini Acked-by: David Hildenbrand Reviewed-by: Oscar Salvador Cc: Alexey Makhalov Cc: Christoph Lameter Cc: Dennis Zhou Cc: Eric Dumazet Cc: Mike Rapoport Cc: Nico Pache Cc: Tejun Heo Cc: Wei Yang Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memory_hotplug.h | 2 +- mm/memory_hotplug.c | 28 +++------------------------- mm/page_alloc.c | 25 +++++++++++++++++++++++-- 3 files changed, 27 insertions(+), 28 deletions(-) (limited to 'include/linux/memory_hotplug.h') diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h index 60f09d3ebb3d..76bf2de86def 100644 --- a/include/linux/memory_hotplug.h +++ b/include/linux/memory_hotplug.h @@ -319,7 +319,7 @@ extern void set_zone_contiguous(struct zone *zone); extern void clear_zone_contiguous(struct zone *zone); #ifdef CONFIG_MEMORY_HOTPLUG -extern void __ref free_area_init_core_hotplug(int nid); +extern void __ref free_area_init_core_hotplug(struct pglist_data *pgdat); extern int __add_memory(int nid, u64 start, u64 size, mhp_t mhp_flags); extern int add_memory(int nid, u64 start, u64 size, mhp_t mhp_flags); extern int add_memory_resource(int nid, struct resource *resource, diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c index 55c3e5309088..a4f69d399929 100644 --- a/mm/memory_hotplug.c +++ b/mm/memory_hotplug.c @@ -1166,39 +1166,16 @@ static pg_data_t __ref *hotadd_init_pgdat(int nid) { struct pglist_data *pgdat; - pgdat = NODE_DATA(nid); - /* * NODE_DATA is preallocated (free_area_init) but its internal * state is not allocated completely. Add missing pieces. * Completely offline nodes stay around and they just need * reintialization. */ - if (pgdat->per_cpu_nodestats == &boot_nodestats) { - pgdat->per_cpu_nodestats = - alloc_percpu(struct per_cpu_nodestat); - } else { - int cpu; - /* - * Reset the nr_zones, order and highest_zoneidx before reuse. - * Note that kswapd will init kswapd_highest_zoneidx properly - * when it starts in the near future. - */ - pgdat->nr_zones = 0; - pgdat->kswapd_order = 0; - pgdat->kswapd_highest_zoneidx = 0; - for_each_online_cpu(cpu) { - struct per_cpu_nodestat *p; - - p = per_cpu_ptr(pgdat->per_cpu_nodestats, cpu); - memset(p, 0, sizeof(*p)); - } - } - - pgdat->node_start_pfn = 0; + pgdat = NODE_DATA(nid); /* init node's zones as empty zones, we don't have any present pages.*/ - free_area_init_core_hotplug(nid); + free_area_init_core_hotplug(pgdat); /* * The node we allocated has no zone fallback lists. For avoiding @@ -1210,6 +1187,7 @@ static pg_data_t __ref *hotadd_init_pgdat(int nid) * When memory is hot-added, all the memory is in offline state. So * clear all zones' present_pages because they will be updated in * online_pages() and offline_pages(). + * TODO: should be in free_area_init_core_hotplug? */ reset_node_managed_pages(pgdat); reset_node_present_pages(pgdat); diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 4f141a4e5b64..d32a635f2c72 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -7466,12 +7466,33 @@ static void __meminit zone_init_internals(struct zone *zone, enum zone_type idx, * NOTE: this function is only called during memory hotplug */ #ifdef CONFIG_MEMORY_HOTPLUG -void __ref free_area_init_core_hotplug(int nid) +void __ref free_area_init_core_hotplug(struct pglist_data *pgdat) { + int nid = pgdat->node_id; enum zone_type z; - pg_data_t *pgdat = NODE_DATA(nid); + int cpu; pgdat_init_internals(pgdat); + + if (pgdat->per_cpu_nodestats == &boot_nodestats) + pgdat->per_cpu_nodestats = alloc_percpu(struct per_cpu_nodestat); + + /* + * Reset the nr_zones, order and highest_zoneidx before reuse. + * Note that kswapd will init kswapd_highest_zoneidx properly + * when it starts in the near future. + */ + pgdat->nr_zones = 0; + pgdat->kswapd_order = 0; + pgdat->kswapd_highest_zoneidx = 0; + pgdat->node_start_pfn = 0; + for_each_online_cpu(cpu) { + struct per_cpu_nodestat *p; + + p = per_cpu_ptr(pgdat->per_cpu_nodestats, cpu); + memset(p, 0, sizeof(*p)); + } + for (z = 0; z < MAX_NR_ZONES; z++) zone_init_internals(&pgdat->node_zones[z], z, nid, 0); } -- cgit v1.2.3-71-gd317 From 395f6081bad49f9c54abafebab49ee23aa985bbd Mon Sep 17 00:00:00 2001 From: David Hildenbrand Date: Tue, 22 Mar 2022 14:47:31 -0700 Subject: drivers/base/memory: determine and store zone for single-zone memory blocks test_pages_in_a_zone() is just another nasty PFN walker that can easily stumble over ZONE_DEVICE memory ranges falling into the same memory block as ordinary system RAM: the memmap of parts of these ranges might possibly be uninitialized. In fact, we observed (on an older kernel) with UBSAN: UBSAN: Undefined behaviour in ./include/linux/mm.h:1133:50 index 7 is out of range for type 'zone [5]' CPU: 121 PID: 35603 Comm: read_all Kdump: loaded Tainted: [...] Hardware name: Dell Inc. PowerEdge R7425/08V001, BIOS 1.12.2 11/15/2019 Call Trace: dump_stack+0x9a/0xf0 ubsan_epilogue+0x9/0x7a __ubsan_handle_out_of_bounds+0x13a/0x181 test_pages_in_a_zone+0x3c4/0x500 show_valid_zones+0x1fa/0x380 dev_attr_show+0x43/0xb0 sysfs_kf_seq_show+0x1c5/0x440 seq_read+0x49d/0x1190 vfs_read+0xff/0x300 ksys_read+0xb8/0x170 do_syscall_64+0xa5/0x4b0 entry_SYSCALL_64_after_hwframe+0x6a/0xdf RIP: 0033:0x7f01f4439b52 We seem to stumble over a memmap that contains a garbage zone id. While we could try inserting pfn_to_online_page() calls, it will just make memory offlining slower, because we use test_pages_in_a_zone() to make sure we're offlining pages that all belong to the same zone. Let's just get rid of this PFN walker and determine the single zone of a memory block -- if any -- for early memory blocks during boot. For memory onlining, we know the single zone already. Let's avoid any additional memmap scanning and just rely on the zone information available during boot. For memory hot(un)plug, we only really care about memory blocks that: * span a single zone (and, thereby, a single node) * are completely System RAM (IOW, no holes, no ZONE_DEVICE) If one of these conditions is not met, we reject memory offlining. Hotplugged memory blocks (starting out offline), always meet both conditions. There are three scenarios to handle: (1) Memory hot(un)plug A memory block with zone == NULL cannot be offlined, corresponding to our previous test_pages_in_a_zone() check. After successful memory onlining/offlining, we simply set the zone accordingly. * Memory onlining: set the zone we just used for onlining * Memory offlining: set zone = NULL So a hotplugged memory block starts with zone = NULL. Once memory onlining is done, we set the proper zone. (2) Boot memory with !CONFIG_NUMA We know that there is just a single pgdat, so we simply scan all zones of that pgdat for an intersection with our memory block PFN range when adding the memory block. If more than one zone intersects (e.g., DMA and DMA32 on x86 for the first memory block) we set zone = NULL and consequently mimic what test_pages_in_a_zone() used to do. (3) Boot memory with CONFIG_NUMA At the point in time we create the memory block devices during boot, we don't know yet which nodes *actually* span a memory block. While we could scan all zones of all nodes for intersections, overlapping nodes complicate the situation and scanning all nodes is possibly expensive. But that problem has already been solved by the code that sets the node of a memory block and creates the link in the sysfs -- do_register_memory_block_under_node(). So, we hook into the code that sets the node id for a memory block. If we already have a different node id set for the memory block, we know that multiple nodes *actually* have PFNs falling into our memory block: we set zone = NULL and consequently mimic what test_pages_in_a_zone() used to do. If there is no node id set, we do the same as (2) for the given node. Note that the call order in driver_init() is: -> memory_dev_init(): create memory block devices -> node_dev_init(): link memory block devices to the node and set the node id So in summary, we detect if there is a single zone responsible for this memory block and we consequently store the zone in that case in the memory block, updating it during memory onlining/offlining. Link: https://lkml.kernel.org/r/20220210184359.235565-3-david@redhat.com Signed-off-by: David Hildenbrand Reported-by: Rafael Parra Reviewed-by: Oscar Salvador Cc: "Rafael J. Wysocki" Cc: Greg Kroah-Hartman Cc: Michal Hocko Cc: Rafael Parra Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- drivers/base/memory.c | 101 +++++++++++++++++++++++++++++++++++++++-- drivers/base/node.c | 13 ++---- include/linux/memory.h | 12 +++++ include/linux/memory_hotplug.h | 6 +-- mm/memory_hotplug.c | 50 ++++---------------- 5 files changed, 125 insertions(+), 57 deletions(-) (limited to 'include/linux/memory_hotplug.h') diff --git a/drivers/base/memory.c b/drivers/base/memory.c index 6ee2181adc3f..f75e3467cb59 100644 --- a/drivers/base/memory.c +++ b/drivers/base/memory.c @@ -215,6 +215,7 @@ static int memory_block_online(struct memory_block *mem) adjust_present_page_count(pfn_to_page(start_pfn), mem->group, nr_vmemmap_pages); + mem->zone = zone; return ret; } @@ -225,6 +226,9 @@ static int memory_block_offline(struct memory_block *mem) unsigned long nr_vmemmap_pages = mem->nr_vmemmap_pages; int ret; + if (!mem->zone) + return -EINVAL; + /* * Unaccount before offlining, such that unpopulated zone and kthreads * can properly be torn down in offline_pages(). @@ -234,7 +238,7 @@ static int memory_block_offline(struct memory_block *mem) -nr_vmemmap_pages); ret = offline_pages(start_pfn + nr_vmemmap_pages, - nr_pages - nr_vmemmap_pages, mem->group); + nr_pages - nr_vmemmap_pages, mem->zone, mem->group); if (ret) { /* offline_pages() failed. Account back. */ if (nr_vmemmap_pages) @@ -246,6 +250,7 @@ static int memory_block_offline(struct memory_block *mem) if (nr_vmemmap_pages) mhp_deinit_memmap_on_memory(start_pfn, nr_vmemmap_pages); + mem->zone = NULL; return ret; } @@ -411,11 +416,10 @@ static ssize_t valid_zones_show(struct device *dev, */ if (mem->state == MEM_ONLINE) { /* - * The block contains more than one zone can not be offlined. - * This can happen e.g. for ZONE_DMA and ZONE_DMA32 + * If !mem->zone, the memory block spans multiple zones and + * cannot get offlined. */ - default_zone = test_pages_in_a_zone(start_pfn, - start_pfn + nr_pages); + default_zone = mem->zone; if (!default_zone) return sysfs_emit(buf, "%s\n", "none"); len += sysfs_emit_at(buf, len, "%s", default_zone->name); @@ -643,6 +647,82 @@ int register_memory(struct memory_block *memory) return ret; } +static struct zone *early_node_zone_for_memory_block(struct memory_block *mem, + int nid) +{ + const unsigned long start_pfn = section_nr_to_pfn(mem->start_section_nr); + const unsigned long nr_pages = PAGES_PER_SECTION * sections_per_block; + struct zone *zone, *matching_zone = NULL; + pg_data_t *pgdat = NODE_DATA(nid); + int i; + + /* + * This logic only works for early memory, when the applicable zones + * already span the memory block. We don't expect overlapping zones on + * a single node for early memory. So if we're told that some PFNs + * of a node fall into this memory block, we can assume that all node + * zones that intersect with the memory block are actually applicable. + * No need to look at the memmap. + */ + for (i = 0; i < MAX_NR_ZONES; i++) { + zone = pgdat->node_zones + i; + if (!populated_zone(zone)) + continue; + if (!zone_intersects(zone, start_pfn, nr_pages)) + continue; + if (!matching_zone) { + matching_zone = zone; + continue; + } + /* Spans multiple zones ... */ + matching_zone = NULL; + break; + } + return matching_zone; +} + +#ifdef CONFIG_NUMA +/** + * memory_block_add_nid() - Indicate that system RAM falling into this memory + * block device (partially) belongs to the given node. + * @mem: The memory block device. + * @nid: The node id. + * @context: The memory initialization context. + * + * Indicate that system RAM falling into this memory block (partially) belongs + * to the given node. If the context indicates ("early") that we are adding the + * node during node device subsystem initialization, this will also properly + * set/adjust mem->zone based on the zone ranges of the given node. + */ +void memory_block_add_nid(struct memory_block *mem, int nid, + enum meminit_context context) +{ + if (context == MEMINIT_EARLY && mem->nid != nid) { + /* + * For early memory we have to determine the zone when setting + * the node id and handle multiple nodes spanning a single + * memory block by indicate via zone == NULL that we're not + * dealing with a single zone. So if we're setting the node id + * the first time, determine if there is a single zone. If we're + * setting the node id a second time to a different node, + * invalidate the single detected zone. + */ + if (mem->nid == NUMA_NO_NODE) + mem->zone = early_node_zone_for_memory_block(mem, nid); + else + mem->zone = NULL; + } + + /* + * If this memory block spans multiple nodes, we only indicate + * the last processed node. If we span multiple nodes (not applicable + * to hotplugged memory), zone == NULL will prohibit memory offlining + * and consequently unplug. + */ + mem->nid = nid; +} +#endif + static int init_memory_block(unsigned long block_id, unsigned long state, unsigned long nr_vmemmap_pages, struct memory_group *group) @@ -665,6 +745,17 @@ static int init_memory_block(unsigned long block_id, unsigned long state, mem->nr_vmemmap_pages = nr_vmemmap_pages; INIT_LIST_HEAD(&mem->group_next); +#ifndef CONFIG_NUMA + if (state == MEM_ONLINE) + /* + * MEM_ONLINE at this point implies early memory. With NUMA, + * we'll determine the zone when setting the node id via + * memory_block_add_nid(). Memory hotplug updated the zone + * manually when memory onlining/offlining succeeds. + */ + mem->zone = early_node_zone_for_memory_block(mem, NUMA_NO_NODE); +#endif /* CONFIG_NUMA */ + ret = register_memory(mem); if (ret) return ret; diff --git a/drivers/base/node.c b/drivers/base/node.c index 5d75341413ce..ec8bb24a5a22 100644 --- a/drivers/base/node.c +++ b/drivers/base/node.c @@ -796,15 +796,12 @@ static int __ref get_nid_for_pfn(unsigned long pfn) } static void do_register_memory_block_under_node(int nid, - struct memory_block *mem_blk) + struct memory_block *mem_blk, + enum meminit_context context) { int ret; - /* - * If this memory block spans multiple nodes, we only indicate - * the last processed node. - */ - mem_blk->nid = nid; + memory_block_add_nid(mem_blk, nid, context); ret = sysfs_create_link_nowarn(&node_devices[nid]->dev.kobj, &mem_blk->dev.kobj, @@ -857,7 +854,7 @@ static int register_mem_block_under_node_early(struct memory_block *mem_blk, if (page_nid != nid) continue; - do_register_memory_block_under_node(nid, mem_blk); + do_register_memory_block_under_node(nid, mem_blk, MEMINIT_EARLY); return 0; } /* mem section does not span the specified node */ @@ -873,7 +870,7 @@ static int register_mem_block_under_node_hotplug(struct memory_block *mem_blk, { int nid = *(int *)arg; - do_register_memory_block_under_node(nid, mem_blk); + do_register_memory_block_under_node(nid, mem_blk, MEMINIT_HOTPLUG); return 0; } diff --git a/include/linux/memory.h b/include/linux/memory.h index 88eb587b5143..aa619464a1df 100644 --- a/include/linux/memory.h +++ b/include/linux/memory.h @@ -70,6 +70,13 @@ struct memory_block { unsigned long state; /* serialized by the dev->lock */ int online_type; /* for passing data to online routine */ int nid; /* NID for this memory block */ + /* + * The single zone of this memory block if all PFNs of this memory block + * that are System RAM (not a memory hole, not ZONE_DEVICE ranges) are + * managed by a single zone. NULL if multiple zones (including nodes) + * apply. + */ + struct zone *zone; struct device dev; /* * Number of vmemmap pages. These pages @@ -161,6 +168,11 @@ int walk_dynamic_memory_groups(int nid, walk_memory_groups_func_t func, }) #define register_hotmemory_notifier(nb) register_memory_notifier(nb) #define unregister_hotmemory_notifier(nb) unregister_memory_notifier(nb) + +#ifdef CONFIG_NUMA +void memory_block_add_nid(struct memory_block *mem, int nid, + enum meminit_context context); +#endif /* CONFIG_NUMA */ #endif /* CONFIG_MEMORY_HOTPLUG */ /* diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h index 76bf2de86def..1ce6f8044f1e 100644 --- a/include/linux/memory_hotplug.h +++ b/include/linux/memory_hotplug.h @@ -163,8 +163,6 @@ extern int mhp_init_memmap_on_memory(unsigned long pfn, unsigned long nr_pages, 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); -extern struct zone *test_pages_in_a_zone(unsigned long start_pfn, - unsigned long end_pfn); extern void __offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn); @@ -293,7 +291,7 @@ static inline void pgdat_resize_init(struct pglist_data *pgdat) {} extern void try_offline_node(int nid); extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages, - struct memory_group *group); + struct zone *zone, struct memory_group *group); extern int remove_memory(u64 start, u64 size); extern void __remove_memory(u64 start, u64 size); extern int offline_and_remove_memory(u64 start, u64 size); @@ -302,7 +300,7 @@ extern int offline_and_remove_memory(u64 start, u64 size); static inline void try_offline_node(int nid) {} static inline int offline_pages(unsigned long start_pfn, unsigned long nr_pages, - struct memory_group *group) + struct zone *zone, struct memory_group *group) { return -EINVAL; } diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c index ed1a5dac6797..aee69281dad6 100644 --- a/mm/memory_hotplug.c +++ b/mm/memory_hotplug.c @@ -1548,38 +1548,6 @@ bool mhp_range_allowed(u64 start, u64 size, bool need_mapping) } #ifdef CONFIG_MEMORY_HOTREMOVE -/* - * Confirm all pages in a range [start, end) belong to the same zone (skipping - * memory holes). When true, return the zone. - */ -struct zone *test_pages_in_a_zone(unsigned long start_pfn, - unsigned long end_pfn) -{ - unsigned long pfn, sec_end_pfn; - struct zone *zone = NULL; - struct page *page; - - for (pfn = start_pfn, sec_end_pfn = SECTION_ALIGN_UP(start_pfn + 1); - pfn < end_pfn; - pfn = sec_end_pfn, sec_end_pfn += PAGES_PER_SECTION) { - /* Make sure the memory section is present first */ - if (!present_section_nr(pfn_to_section_nr(pfn))) - continue; - for (; pfn < sec_end_pfn && pfn < end_pfn; - pfn += MAX_ORDER_NR_PAGES) { - /* Check if we got outside of the zone */ - if (zone && !zone_spans_pfn(zone, pfn)) - return NULL; - page = pfn_to_page(pfn); - if (zone && page_zone(page) != zone) - return NULL; - zone = page_zone(page); - } - } - - return zone; -} - /* * Scan pfn range [start,end) to find movable/migratable pages (LRU pages, * non-lru movable pages and hugepages). Will skip over most unmovable @@ -1803,15 +1771,15 @@ static int count_system_ram_pages_cb(unsigned long start_pfn, } int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages, - struct memory_group *group) + struct zone *zone, struct memory_group *group) { const unsigned long end_pfn = start_pfn + nr_pages; unsigned long pfn, system_ram_pages = 0; + const int node = zone_to_nid(zone); unsigned long flags; - struct zone *zone; struct memory_notify arg; - int ret, node; char *reason; + int ret; /* * {on,off}lining is constrained to full memory sections (or more @@ -1843,15 +1811,17 @@ int __ref offline_pages(unsigned long start_pfn, unsigned long nr_pages, goto failed_removal; } - /* This makes hotplug much easier...and readable. - we assume this for now. .*/ - zone = test_pages_in_a_zone(start_pfn, end_pfn); - if (!zone) { + /* + * We only support offlining of memory blocks managed by a single zone, + * checked by calling code. This is just a sanity check that we might + * want to remove in the future. + */ + if (WARN_ON_ONCE(page_zone(pfn_to_page(start_pfn)) != zone || + page_zone(pfn_to_page(end_pfn - 1)) != zone)) { ret = -EINVAL; reason = "multizone range"; goto failed_removal; } - node = zone_to_nid(zone); /* * Disable pcplists so that page isolation cannot race with freeing -- cgit v1.2.3-71-gd317