mm: meminit: finish initialisation of struct pages before basic setup

Waiman Long reported that 24TB machines hit OOM during basic setup when
struct page initialisation was deferred.  One approach is to initialise
memory on demand but it interferes with page allocator paths.  This patch
creates dedicated threads to initialise memory before basic setup.  It
then blocks on a rw_semaphore until completion as a wait_queue and counter
is overkill.  This may be slower to boot but it's simplier overall and
also gets rid of a section mangling which existed so kswapd could do the
initialisation.

[[email protected]: include rwsem.h, use DECLARE_RWSEM, fix comment, remove unneeded cast]
Signed-off-by: Mel Gorman <[email protected]>
Cc: Waiman Long <[email protected]
Cc: Nathan Zimmer <[email protected]>
Cc: Dave Hansen <[email protected]>
Cc: Scott Norton <[email protected]>
Tested-by: Daniel J Blueman <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

include/linux/gfp.h

@@ -384,6 +384,14 @@ void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp);
 void drain_all_pages(struct zone *zone);
 void drain_local_pages(struct zone *zone);
 
+#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
+void page_alloc_init_late(void);
+#else
+static inline void page_alloc_init_late(void)
+{
+}
+#endif
+
 /*
  * gfp_allowed_mask is set to GFP_BOOT_MASK during early boot to restrict what
  * GFP flags are used before interrupts are enabled. Once interrupts are

init/main.c

@@ -1004,6 +1004,8 @@ static noinline void __init kernel_init_freeable(void)
 	smp_init();
 	sched_init_smp();
 
+	page_alloc_init_late();
+
 	do_basic_setup();
 
 	/* Open the /dev/console on the rootfs, this should never fail */

mm/internal.h

@@ -379,30 +379,6 @@ static inline void mminit_verify_zonelist(void)
 }
 #endif /* CONFIG_DEBUG_MEMORY_INIT */
 
-/*
- * Deferred struct page initialisation requires init functions that are freed
- * before kswapd is available. Reuse the memory hotplug section annotation
- * to mark the required code.
- *
- * __defermem_init is code that always exists but is annotated __meminit to
- * 	avoid section warnings.
- * __defer_init code gets marked __meminit when deferring struct page
- *	initialistion but is otherwise in the init section.
- */
-#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
-#define __defermem_init __meminit
-#define __defer_init    __meminit
-
-void deferred_init_memmap(int nid);
-#else
-#define __defermem_init
-#define __defer_init __init
-
-static inline void deferred_init_memmap(int nid)
-{
-}
-#endif
-
 /* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */
 #if defined(CONFIG_SPARSEMEM)
 extern void mminit_validate_memmodel_limits(unsigned long *start_pfn,

mm/page_alloc.c

@@ -18,6 +18,7 @@
 #include <linux/mm.h>
 #include <linux/swap.h>
 #include <linux/interrupt.h>
+#include <linux/rwsem.h>
 #include <linux/pagemap.h>
 #include <linux/jiffies.h>
 #include <linux/bootmem.h>
@@ -61,6 +62,7 @@
 #include <linux/hugetlb.h>
 #include <linux/sched/rt.h>
 #include <linux/page_owner.h>
+#include <linux/kthread.h>
 
 #include <asm/sections.h>
 #include <asm/tlbflush.h>
@@ -242,7 +244,7 @@ static inline void reset_deferred_meminit(pg_data_t *pgdat)
 }
 
 /* Returns true if the struct page for the pfn is uninitialised */
-static inline bool __defermem_init early_page_uninitialised(unsigned long pfn)
+static inline bool __meminit early_page_uninitialised(unsigned long pfn)
 {
 	int nid = early_pfn_to_nid(pfn);
 
@@ -958,7 +960,7 @@ static void __free_pages_ok(struct page *page, unsigned int order)
 	local_irq_restore(flags);
 }
 
-static void __defer_init __free_pages_boot_core(struct page *page,
+static void __init __free_pages_boot_core(struct page *page,
 					unsigned long pfn, unsigned int order)
 {
 	unsigned int nr_pages = 1 << order;
@@ -1031,7 +1033,7 @@ static inline bool __meminit meminit_pfn_in_nid(unsigned long pfn, int node,
 #endif
 
 
-void __defer_init __free_pages_bootmem(struct page *page, unsigned long pfn,
+void __init __free_pages_bootmem(struct page *page, unsigned long pfn,
 							unsigned int order)
 {
 	if (early_page_uninitialised(pfn))
@@ -1040,7 +1042,7 @@ void __defer_init __free_pages_bootmem(struct page *page, unsigned long pfn,
 }
 
 #ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
-static void __defermem_init deferred_free_range(struct page *page,
+static void __init deferred_free_range(struct page *page,
 					unsigned long pfn, int nr_pages)
 {
 	int i;
@@ -1060,20 +1062,30 @@ static void __defermem_init deferred_free_range(struct page *page,
 		__free_pages_boot_core(page, pfn, 0);
 }
 
+static __initdata DECLARE_RWSEM(pgdat_init_rwsem);
+
 /* Initialise remaining memory on a node */
-void __defermem_init deferred_init_memmap(int nid)
+static int __init deferred_init_memmap(void *data)
 {
+	pg_data_t *pgdat = data;
+	int nid = pgdat->node_id;
 	struct mminit_pfnnid_cache nid_init_state = { };
 	unsigned long start = jiffies;
 	unsigned long nr_pages = 0;
 	unsigned long walk_start, walk_end;
 	int i, zid;
 	struct zone *zone;
-	pg_data_t *pgdat = NODE_DATA(nid);
 	unsigned long first_init_pfn = pgdat->first_deferred_pfn;
+	const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id);
 
-	if (first_init_pfn == ULONG_MAX)
-		return;
+	if (first_init_pfn == ULONG_MAX) {
+		up_read(&pgdat_init_rwsem);
+		return 0;
+	}
+
+	/* Bind memory initialisation thread to a local node if possible */
+	if (!cpumask_empty(cpumask))
+		set_cpus_allowed_ptr(current, cpumask);
 
 	/* Sanity check boundaries */
 	BUG_ON(pgdat->first_deferred_pfn < pgdat->node_start_pfn);
@@ -1165,8 +1177,24 @@ void __defermem_init deferred_init_memmap(int nid)
 	/* Sanity check that the next zone really is unpopulated */
 	WARN_ON(++zid < MAX_NR_ZONES && populated_zone(++zone));
 
-	pr_info("kswapd %d initialised %lu pages in %ums\n", nid, nr_pages,
+	pr_info("node %d initialised, %lu pages in %ums\n", nid, nr_pages,
 					jiffies_to_msecs(jiffies - start));
+	up_read(&pgdat_init_rwsem);
+	return 0;
+}
+
+void __init page_alloc_init_late(void)
+{
+	int nid;
+
+	for_each_node_state(nid, N_MEMORY) {
+		down_read(&pgdat_init_rwsem);
+		kthread_run(deferred_init_memmap, NODE_DATA(nid), "pgdatinit%d", nid);
+	}
+
+	/* Block until all are initialised */
+	down_write(&pgdat_init_rwsem);
+	up_write(&pgdat_init_rwsem);
 }
 #endif /* CONFIG_DEFERRED_STRUCT_PAGE_INIT */
 

mm/vmscan.c

@@ -3386,7 +3386,7 @@ static void kswapd_try_to_sleep(pg_data_t *pgdat, int order, int classzone_idx)
  * If there are applications that are active memory-allocators
  * (most normal use), this basically shouldn't matter.
  */
-static int __defermem_init kswapd(void *p)
+static int kswapd(void *p)
 {
 	unsigned long order, new_order;
 	unsigned balanced_order;
@@ -3421,8 +3421,6 @@ static int __defermem_init kswapd(void *p)
 	tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
 	set_freezable();
 
-	deferred_init_memmap(pgdat->node_id);
-
 	order = new_order = 0;
 	balanced_order = 0;
 	classzone_idx = new_classzone_idx = pgdat->nr_zones - 1;
@@ -3578,7 +3576,7 @@ static int cpu_callback(struct notifier_block *nfb, unsigned long action,
  * This kswapd start function will be called by init and node-hot-add.
  * On node-hot-add, kswapd will moved to proper cpus if cpus are hot-added.
  */
-int __defermem_init kswapd_run(int nid)
+int kswapd_run(int nid)
 {
 	pg_data_t *pgdat = NODE_DATA(nid);
 	int ret = 0;