diff options
Diffstat (limited to 'mm')
| -rw-r--r-- | mm/oom_kill.c | 23 | ||||
| -rw-r--r-- | mm/show_mem.c | 50 | ||||
| -rw-r--r-- | mm/shrinker.c | 107 | ||||
| -rw-r--r-- | mm/shrinker_debug.c | 18 | ||||
| -rw-r--r-- | mm/slab.h | 6 | ||||
| -rw-r--r-- | mm/slab_common.c | 52 |
6 files changed, 221 insertions, 35 deletions
diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 25923cfec9c6..ad1ac9be0db4 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -169,27 +169,6 @@ static bool oom_unkillable_task(struct task_struct *p) return false; } -/* - * Check whether unreclaimable slab amount is greater than - * all user memory(LRU pages). - * dump_unreclaimable_slab() could help in the case that - * oom due to too much unreclaimable slab used by kernel. -*/ -static bool should_dump_unreclaim_slab(void) -{ - unsigned long nr_lru; - - nr_lru = global_node_page_state(NR_ACTIVE_ANON) + - global_node_page_state(NR_INACTIVE_ANON) + - global_node_page_state(NR_ACTIVE_FILE) + - global_node_page_state(NR_INACTIVE_FILE) + - global_node_page_state(NR_ISOLATED_ANON) + - global_node_page_state(NR_ISOLATED_FILE) + - global_node_page_state(NR_UNEVICTABLE); - - return (global_node_page_state_pages(NR_SLAB_UNRECLAIMABLE_B) > nr_lru); -} - /** * oom_badness - heuristic function to determine which candidate task to kill * @p: task struct of which task we should calculate @@ -469,8 +448,6 @@ static void dump_header(struct oom_control *oc) mem_cgroup_print_oom_meminfo(oc->memcg); else { __show_mem(SHOW_MEM_FILTER_NODES, oc->nodemask, gfp_zone(oc->gfp_mask)); - if (should_dump_unreclaim_slab()) - dump_unreclaimable_slab(); } if (sysctl_oom_dump_tasks) dump_tasks(oc); diff --git a/mm/show_mem.c b/mm/show_mem.c index 41999e94a56d..013ad4da618c 100644 --- a/mm/show_mem.c +++ b/mm/show_mem.c @@ -7,15 +7,18 @@ #include <linux/blkdev.h> #include <linux/cma.h> +#include <linux/console.h> #include <linux/cpuset.h> #include <linux/highmem.h> #include <linux/hugetlb.h> #include <linux/mm.h> #include <linux/mmzone.h> +#include <linux/seq_buf.h> #include <linux/swap.h> #include <linux/vmstat.h> #include "internal.h" +#include "slab.h" #include "swap.h" atomic_long_t _totalram_pages __read_mostly; @@ -392,10 +395,31 @@ static void show_free_areas(unsigned int filter, nodemask_t *nodemask, int max_z show_swap_cache_info(); } +static void print_string_as_lines(const char *prefix, const char *lines) +{ + if (!lines) { + printk("%s (null)\n", prefix); + return; + } + + bool locked = console_trylock(); + + while (1) { + const char *p = strchrnul(lines, '\n'); + printk("%s%.*s\n", prefix, (int) (p - lines), lines); + if (!*p) + break; + lines = p + 1; + } + if (locked) + console_unlock(); +} + void __show_mem(unsigned int filter, nodemask_t *nodemask, int max_zone_idx) { unsigned long total = 0, reserved = 0, highmem = 0; struct zone *zone; + char *buf; printk("Mem-Info:\n"); show_free_areas(filter, nodemask, max_zone_idx); @@ -447,4 +471,30 @@ void __show_mem(unsigned int filter, nodemask_t *nodemask, int max_zone_idx) } } #endif + + const unsigned buf_size = 8192; + buf = kmalloc(buf_size, GFP_ATOMIC); + if (buf) { + struct seq_buf s; + + printk("Unreclaimable slab info:\n"); + seq_buf_init(&s, buf, buf_size); + dump_unreclaimable_slab(&s); + print_string_as_lines(KERN_NOTICE, seq_buf_str(&s)); + + static unsigned long shrinkers_last_print; + + /* Ratelimit to at most once every 30 seconds */ + if (!shrinkers_last_print || + time_after(jiffies, shrinkers_last_print + HZ * 30)) { + shrinkers_last_print = jiffies; + + printk("Shrinkers:\n"); + seq_buf_init(&s, buf, buf_size); + shrinkers_to_text(&s); + print_string_as_lines(KERN_NOTICE, seq_buf_str(&s)); + } + + kfree(buf); + } } diff --git a/mm/shrinker.c b/mm/shrinker.c index 4a93fd433689..4a76364d2b7e 100644 --- a/mm/shrinker.c +++ b/mm/shrinker.c @@ -1,8 +1,9 @@ // SPDX-License-Identifier: GPL-2.0 #include <linux/memcontrol.h> +#include <linux/rculist.h> #include <linux/rwsem.h> +#include <linux/seq_buf.h> #include <linux/shrinker.h> -#include <linux/rculist.h> #include <trace/events/vmscan.h> #include "internal.h" @@ -411,6 +412,7 @@ static unsigned long do_shrink_slab(struct shrink_control *shrinkctl, trace_mm_shrink_slab_start(shrinker, shrinkctl, nr, freeable, delta, total_scan, priority); + u64 start_time = ktime_get_ns(); /* * Normally, we should not scan less than batch_size objects in one @@ -461,6 +463,17 @@ static unsigned long do_shrink_slab(struct shrink_control *shrinkctl, */ new_nr = add_nr_deferred(next_deferred, shrinker, shrinkctl); + unsigned long now = jiffies; + if (freed) { + atomic_long_add(freed, &shrinker->objects_freed); + shrinker->last_freed = now; + } + shrinker->last_scanned = now; + atomic_long_add(scanned, &shrinker->objects_requested_to_free); + + atomic64_add(ktime_get_ns() - start_time, &shrinker->ns_run); + + trace_mm_shrink_slab_end(shrinker, shrinkctl->nid, freed, nr, new_nr, total_scan); return freed; } @@ -809,3 +822,95 @@ void shrinker_free(struct shrinker *shrinker) call_rcu(&shrinker->rcu, shrinker_free_rcu_cb); } EXPORT_SYMBOL_GPL(shrinker_free); + +void shrinker_to_text(struct seq_buf *out, struct shrinker *shrinker) +{ + struct shrink_control sc = { + .gfp_mask = GFP_KERNEL, +#ifdef CONFIG_MEMCG + .memcg = root_mem_cgroup, +#endif + }; + unsigned long nr_freed = atomic_long_read(&shrinker->objects_freed); + + seq_buf_printf(out, "%ps", shrinker->scan_objects); + if (shrinker->name) + seq_buf_printf(out, ": %s", shrinker->name); + seq_buf_putc(out, '\n'); + + seq_buf_printf(out, "objects: %lu\n", shrinker->count_objects(shrinker, &sc)); + seq_buf_printf(out, "requested to free: %lu\n", atomic_long_read(&shrinker->objects_requested_to_free)); + seq_buf_printf(out, "objects freed: %lu\n", nr_freed); + seq_buf_printf(out, "last scanned: %li sec ago\n", (jiffies - shrinker->last_scanned) / HZ); + seq_buf_printf(out, "last freed: %li sec ago\n", (jiffies - shrinker->last_freed) / HZ); + seq_buf_printf(out, "ns per object freed: %llu\n", nr_freed + ? div64_ul(atomic64_read(&shrinker->ns_run), nr_freed) + : 0); + + if (shrinker->to_text) { + shrinker->to_text(out, shrinker); + seq_buf_puts(out, "\n"); + } +} + +/** + * shrinkers_to_text - Report on shrinkers with highest usage + * + * This reports on the top 10 shrinkers, by object counts, in sorted order: + * intended to be used for OOM reporting. + */ +void shrinkers_to_text(struct seq_buf *out) +{ + struct shrinker *shrinker; + struct shrinker_by_mem { + struct shrinker *shrinker; + unsigned long mem; + } shrinkers_by_mem[4]; + int i, nr = 0; + + if (!mutex_trylock(&shrinker_mutex)) { + seq_buf_puts(out, "(couldn't take shrinker lock)"); + return; + } + + list_for_each_entry(shrinker, &shrinker_list, list) { + struct shrink_control sc = { + .gfp_mask = GFP_KERNEL, +#ifdef CONFIG_MEMCG + .memcg = root_mem_cgroup, +#endif + }; + unsigned long mem = shrinker->count_objects(shrinker, &sc); + + if (!mem || mem == SHRINK_STOP || mem == SHRINK_EMPTY) + continue; + + for (i = 0; i < nr; i++) + if (mem < shrinkers_by_mem[i].mem) + break; + + if (nr < ARRAY_SIZE(shrinkers_by_mem)) { + memmove(&shrinkers_by_mem[i + 1], + &shrinkers_by_mem[i], + sizeof(shrinkers_by_mem[0]) * (nr - i)); + nr++; + } else if (i) { + i--; + memmove(&shrinkers_by_mem[0], + &shrinkers_by_mem[1], + sizeof(shrinkers_by_mem[0]) * i); + } else { + continue; + } + + shrinkers_by_mem[i] = (struct shrinker_by_mem) { + .shrinker = shrinker, + .mem = mem, + }; + } + + for (i = nr - 1; i >= 0; --i) + shrinker_to_text(out, shrinkers_by_mem[i].shrinker); + + mutex_unlock(&shrinker_mutex); +} diff --git a/mm/shrinker_debug.c b/mm/shrinker_debug.c index 20eaee3e97f7..a16c2848d332 100644 --- a/mm/shrinker_debug.c +++ b/mm/shrinker_debug.c @@ -2,6 +2,7 @@ #include <linux/idr.h> #include <linux/slab.h> #include <linux/debugfs.h> +#include <linux/seq_buf.h> #include <linux/seq_file.h> #include <linux/shrinker.h> #include <linux/memcontrol.h> @@ -159,6 +160,21 @@ static const struct file_operations shrinker_debugfs_scan_fops = { .write = shrinker_debugfs_scan_write, }; +static int shrinker_debugfs_report_show(struct seq_file *m, void *v) +{ + struct shrinker *shrinker = m->private; + char *bufp; + size_t buflen = seq_get_buf(m, &bufp); + struct seq_buf out; + + seq_buf_init(&out, bufp, buflen); + shrinker_to_text(&out, shrinker); + seq_commit(m, seq_buf_used(&out)); + + return 0; +} +DEFINE_SHOW_ATTRIBUTE(shrinker_debugfs_report); + int shrinker_debugfs_add(struct shrinker *shrinker) { struct dentry *entry; @@ -190,6 +206,8 @@ int shrinker_debugfs_add(struct shrinker *shrinker) &shrinker_debugfs_count_fops); debugfs_create_file("scan", 0220, entry, shrinker, &shrinker_debugfs_scan_fops); + debugfs_create_file("report", 0440, entry, shrinker, + &shrinker_debugfs_report_fops); return 0; } diff --git a/mm/slab.h b/mm/slab.h index 248b34c839b7..09be83786b68 100644 --- a/mm/slab.h +++ b/mm/slab.h @@ -587,10 +587,12 @@ static inline size_t slab_ksize(const struct kmem_cache *s) return s->size; } +struct seq_buf; + #ifdef CONFIG_SLUB_DEBUG -void dump_unreclaimable_slab(void); +void dump_unreclaimable_slab(struct seq_buf *); #else -static inline void dump_unreclaimable_slab(void) +static inline void dump_unreclaimable_slab(struct seq_buf *out) { } #endif diff --git a/mm/slab_common.c b/mm/slab_common.c index bfe7c40eeee1..a365f146e669 100644 --- a/mm/slab_common.c +++ b/mm/slab_common.c @@ -27,6 +27,7 @@ #include <asm/tlbflush.h> #include <asm/page.h> #include <linux/memcontrol.h> +#include <linux/seq_buf.h> #include <linux/stackdepot.h> #include <trace/events/rcu.h> @@ -1127,10 +1128,15 @@ static int slab_show(struct seq_file *m, void *p) return 0; } -void dump_unreclaimable_slab(void) +void dump_unreclaimable_slab(struct seq_buf *out) { struct kmem_cache *s; struct slabinfo sinfo; + struct slab_by_mem { + struct kmem_cache *s; + size_t total, active; + } slabs_by_mem[10], n; + int i, nr = 0; /* * Here acquiring slab_mutex is risky since we don't prefer to get @@ -1140,24 +1146,52 @@ void dump_unreclaimable_slab(void) * without acquiring the mutex. */ if (!mutex_trylock(&slab_mutex)) { - pr_warn("excessive unreclaimable slab but cannot dump stats\n"); + seq_buf_puts(out, "excessive unreclaimable slab but cannot dump stats\n"); return; } - pr_info("Unreclaimable slab info:\n"); - pr_info("Name Used Total\n"); - list_for_each_entry(s, &slab_caches, list) { if (s->flags & SLAB_RECLAIM_ACCOUNT) continue; get_slabinfo(s, &sinfo); - if (sinfo.num_objs > 0) - pr_info("%-17s %10luKB %10luKB\n", s->name, - (sinfo.active_objs * s->size) / 1024, - (sinfo.num_objs * s->size) / 1024); + if (!sinfo.num_objs) + continue; + + n.s = s; + n.total = sinfo.num_objs * s->size; + n.active = sinfo.active_objs * s->size; + + for (i = 0; i < nr; i++) + if (n.total < slabs_by_mem[i].total) + break; + + if (nr < ARRAY_SIZE(slabs_by_mem)) { + memmove(&slabs_by_mem[i + 1], + &slabs_by_mem[i], + sizeof(slabs_by_mem[0]) * (nr - i)); + nr++; + } else if (i) { + i--; + memmove(&slabs_by_mem[0], + &slabs_by_mem[1], + sizeof(slabs_by_mem[0]) * i); + } else { + continue; + } + + slabs_by_mem[i] = n; } + + for (i = nr - 1; i >= 0; --i) { + seq_buf_printf(out, "%-17s total: ", slabs_by_mem[i].s->name); + seq_buf_human_readable_u64(out, slabs_by_mem[i].total, STRING_UNITS_2); + seq_buf_printf(out, " active: "); + seq_buf_human_readable_u64(out, slabs_by_mem[i].active, STRING_UNITS_2); + seq_buf_putc(out, '\n'); + } + mutex_unlock(&slab_mutex); } |