diff options
Diffstat (limited to 'src/oom/oomd-util.c')
-rw-r--r-- | src/oom/oomd-util.c | 648 |
1 files changed, 648 insertions, 0 deletions
diff --git a/src/oom/oomd-util.c b/src/oom/oomd-util.c new file mode 100644 index 0000000..f9f0af2 --- /dev/null +++ b/src/oom/oomd-util.c @@ -0,0 +1,648 @@ +/* SPDX-License-Identifier: LGPL-2.1-or-later */ + +#include <sys/xattr.h> +#include <unistd.h> + +#include "errno-util.h" +#include "fd-util.h" +#include "fileio.h" +#include "format-util.h" +#include "memstream-util.h" +#include "oomd-util.h" +#include "parse-util.h" +#include "path-util.h" +#include "procfs-util.h" +#include "signal-util.h" +#include "sort-util.h" +#include "stat-util.h" +#include "stdio-util.h" +#include "user-util.h" + +DEFINE_HASH_OPS_WITH_VALUE_DESTRUCTOR( + oomd_cgroup_ctx_hash_ops, + char, + string_hash_func, + string_compare_func, + OomdCGroupContext, + oomd_cgroup_context_free); + +static int log_kill(const PidRef *pid, int sig, void *userdata) { + log_debug("oomd attempting to kill " PID_FMT " with %s", pid->pid, signal_to_string(sig)); + return 0; +} + +static int increment_oomd_xattr(const char *path, const char *xattr, uint64_t num_procs_killed) { + _cleanup_free_ char *value = NULL; + char buf[DECIMAL_STR_MAX(uint64_t) + 1]; + uint64_t curr_count = 0; + int r; + + assert(path); + assert(xattr); + + r = cg_get_xattr_malloc(path, xattr, &value); + if (r < 0 && !ERRNO_IS_XATTR_ABSENT(r)) + return r; + + if (!isempty(value)) { + r = safe_atou64(value, &curr_count); + if (r < 0) + return r; + } + + if (curr_count > UINT64_MAX - num_procs_killed) + return -EOVERFLOW; + + xsprintf(buf, "%"PRIu64, curr_count + num_procs_killed); + r = cg_set_xattr(path, xattr, buf, strlen(buf), 0); + if (r < 0) + return r; + + return 0; +} + +OomdCGroupContext *oomd_cgroup_context_free(OomdCGroupContext *ctx) { + if (!ctx) + return NULL; + + free(ctx->path); + return mfree(ctx); +} + +int oomd_pressure_above(Hashmap *h, usec_t duration, Set **ret) { + _cleanup_set_free_ Set *targets = NULL; + OomdCGroupContext *ctx; + char *key; + int r; + + assert(h); + assert(ret); + + targets = set_new(NULL); + if (!targets) + return -ENOMEM; + + HASHMAP_FOREACH_KEY(ctx, key, h) { + if (ctx->memory_pressure.avg10 > ctx->mem_pressure_limit) { + usec_t diff; + + if (ctx->mem_pressure_limit_hit_start == 0) + ctx->mem_pressure_limit_hit_start = now(CLOCK_MONOTONIC); + + diff = now(CLOCK_MONOTONIC) - ctx->mem_pressure_limit_hit_start; + if (diff >= duration) { + r = set_put(targets, ctx); + if (r < 0) + return -ENOMEM; + } + } else + ctx->mem_pressure_limit_hit_start = 0; + } + + if (!set_isempty(targets)) { + *ret = TAKE_PTR(targets); + return 1; + } + + *ret = NULL; + return 0; +} + +uint64_t oomd_pgscan_rate(const OomdCGroupContext *c) { + uint64_t last_pgscan; + + assert(c); + + /* If last_pgscan > pgscan, assume the cgroup was recreated and reset last_pgscan to zero. + * pgscan is monotonic and in practice should not decrease (except in the recreation case). */ + last_pgscan = c->last_pgscan; + if (c->last_pgscan > c->pgscan) { + log_debug("Last pgscan %"PRIu64" greater than current pgscan %"PRIu64" for %s. Using last pgscan of zero.", + c->last_pgscan, c->pgscan, c->path); + last_pgscan = 0; + } + + return c->pgscan - last_pgscan; +} + +bool oomd_mem_available_below(const OomdSystemContext *ctx, int threshold_permyriad) { + uint64_t mem_threshold; + + assert(ctx); + assert(threshold_permyriad <= 10000); + + mem_threshold = ctx->mem_total * threshold_permyriad / (uint64_t) 10000; + return LESS_BY(ctx->mem_total, ctx->mem_used) < mem_threshold; +} + +bool oomd_swap_free_below(const OomdSystemContext *ctx, int threshold_permyriad) { + uint64_t swap_threshold; + + assert(ctx); + assert(threshold_permyriad <= 10000); + + swap_threshold = ctx->swap_total * threshold_permyriad / (uint64_t) 10000; + return (ctx->swap_total - ctx->swap_used) < swap_threshold; +} + +int oomd_fetch_cgroup_oom_preference(OomdCGroupContext *ctx, const char *prefix) { + uid_t uid; + int r; + + assert(ctx); + + prefix = empty_to_root(prefix); + + if (!path_startswith(ctx->path, prefix)) + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), + "%s is not a descendant of %s", ctx->path, prefix); + + r = cg_get_owner(ctx->path, &uid); + if (r < 0) + return log_debug_errno(r, "Failed to get owner/group from %s: %m", ctx->path); + + if (uid != 0) { + uid_t prefix_uid; + + r = cg_get_owner(prefix, &prefix_uid); + if (r < 0) + return log_debug_errno(r, "Failed to get owner/group from %s: %m", prefix); + + if (uid != prefix_uid) { + ctx->preference = MANAGED_OOM_PREFERENCE_NONE; + return 0; + } + } + + /* Ignore most errors when reading the xattr since it is usually unset and cgroup xattrs are only used + * as an optional feature of systemd-oomd (and the system might not even support them). */ + r = cg_get_xattr_bool(ctx->path, "user.oomd_avoid"); + if (r == -ENOMEM) + return log_oom_debug(); + if (r < 0 && !ERRNO_IS_XATTR_ABSENT(r)) + log_debug_errno(r, "Failed to get xattr user.oomd_avoid, ignoring: %m"); + ctx->preference = r > 0 ? MANAGED_OOM_PREFERENCE_AVOID : ctx->preference; + + r = cg_get_xattr_bool(ctx->path, "user.oomd_omit"); + if (r == -ENOMEM) + return log_oom_debug(); + if (r < 0 && !ERRNO_IS_XATTR_ABSENT(r)) + log_debug_errno(r, "Failed to get xattr user.oomd_omit, ignoring: %m"); + ctx->preference = r > 0 ? MANAGED_OOM_PREFERENCE_OMIT : ctx->preference; + + return 0; +} + +int oomd_sort_cgroup_contexts(Hashmap *h, oomd_compare_t compare_func, const char *prefix, OomdCGroupContext ***ret) { + _cleanup_free_ OomdCGroupContext **sorted = NULL; + OomdCGroupContext *item; + size_t k = 0; + int r; + + assert(h); + assert(compare_func); + assert(ret); + + sorted = new0(OomdCGroupContext*, hashmap_size(h)); + if (!sorted) + return -ENOMEM; + + HASHMAP_FOREACH(item, h) { + /* Skip over cgroups that are not valid candidates or are explicitly marked for omission */ + if (item->path && prefix && !path_startswith(item->path, prefix)) + continue; + + r = oomd_fetch_cgroup_oom_preference(item, prefix); + if (r == -ENOMEM) + return r; + + if (item->preference == MANAGED_OOM_PREFERENCE_OMIT) + continue; + + sorted[k++] = item; + } + + typesafe_qsort(sorted, k, compare_func); + + *ret = TAKE_PTR(sorted); + + assert(k <= INT_MAX); + return (int) k; +} + +int oomd_cgroup_kill(const char *path, bool recurse, bool dry_run) { + _cleanup_set_free_ Set *pids_killed = NULL; + int r; + + assert(path); + + if (dry_run) { + _cleanup_free_ char *cg_path = NULL; + + r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, path, NULL, &cg_path); + if (r < 0) + return r; + + log_info("oomd dry-run: Would have tried to kill %s with recurse=%s", cg_path, true_false(recurse)); + return 0; + } + + pids_killed = set_new(NULL); + if (!pids_killed) + return -ENOMEM; + + r = increment_oomd_xattr(path, "user.oomd_ooms", 1); + if (r < 0) + log_debug_errno(r, "Failed to set user.oomd_ooms before kill: %m"); + + if (recurse) + r = cg_kill_recursive(path, SIGKILL, CGROUP_IGNORE_SELF, pids_killed, log_kill, NULL); + else + r = cg_kill(path, SIGKILL, CGROUP_IGNORE_SELF, pids_killed, log_kill, NULL); + + /* The cgroup could have been cleaned up after we have sent SIGKILL to all of the processes, but before + * we could do one last iteration of cgroup.procs to check. Or the service unit could have exited and + * was removed between picking candidates and coming into this function. In either case, let's log + * about it let the caller decide what to do once they know how many PIDs were killed. */ + if (IN_SET(r, -ENOENT, -ENODEV)) + log_debug_errno(r, "Error when sending SIGKILL to processes in cgroup path %s, ignoring: %m", path); + else if (r < 0) + return r; + + if (set_isempty(pids_killed)) + log_debug("Nothing killed when attempting to kill %s", path); + + r = increment_oomd_xattr(path, "user.oomd_kill", set_size(pids_killed)); + if (r < 0) + log_debug_errno(r, "Failed to set user.oomd_kill on kill: %m"); + + return set_size(pids_killed) != 0; +} + +typedef void (*dump_candidate_func)(const OomdCGroupContext *ctx, FILE *f, const char *prefix); + +static int dump_kill_candidates(OomdCGroupContext **sorted, int n, int dump_until, dump_candidate_func dump_func) { + _cleanup_(memstream_done) MemStream m = {}; + FILE *f; + + /* Try dumping top offendors, ignoring any errors that might happen. */ + + f = memstream_init(&m); + if (!f) + return -ENOMEM; + + fprintf(f, "Considered %d cgroups for killing, top candidates were:\n", n); + for (int i = 0; i < dump_until; i++) + dump_func(sorted[i], f, "\t"); + + return memstream_dump(LOG_INFO, &m); +} + +int oomd_kill_by_pgscan_rate(Hashmap *h, const char *prefix, bool dry_run, char **ret_selected) { + _cleanup_free_ OomdCGroupContext **sorted = NULL; + int n, r, ret = 0; + int dump_until; + + assert(h); + assert(ret_selected); + + n = oomd_sort_cgroup_contexts(h, compare_pgscan_rate_and_memory_usage, prefix, &sorted); + if (n < 0) + return n; + + dump_until = MIN(n, DUMP_ON_KILL_COUNT); + for (int i = 0; i < n; i++) { + /* Skip cgroups with no reclaim and memory usage; it won't alleviate pressure. + * Continue since there might be "avoid" cgroups at the end. */ + if (sorted[i]->pgscan == 0 && sorted[i]->current_memory_usage == 0) + continue; + + r = oomd_cgroup_kill(sorted[i]->path, /* recurse= */ true, /* dry_run= */ dry_run); + if (r == -ENOMEM) + return r; /* Treat oom as a hard error */ + if (r < 0) { + if (ret == 0) + ret = r; + continue; /* Try to find something else to kill */ + } + + dump_until = MAX(dump_until, i + 1); + char *selected = strdup(sorted[i]->path); + if (!selected) + return -ENOMEM; + *ret_selected = selected; + ret = r; + break; + } + + dump_kill_candidates(sorted, n, dump_until, oomd_dump_memory_pressure_cgroup_context); + + return ret; +} + +int oomd_kill_by_swap_usage(Hashmap *h, uint64_t threshold_usage, bool dry_run, char **ret_selected) { + _cleanup_free_ OomdCGroupContext **sorted = NULL; + int n, r, ret = 0; + int dump_until; + + assert(h); + assert(ret_selected); + + n = oomd_sort_cgroup_contexts(h, compare_swap_usage, NULL, &sorted); + if (n < 0) + return n; + + dump_until = MIN(n, DUMP_ON_KILL_COUNT); + /* Try to kill cgroups with non-zero swap usage until we either succeed in killing or we get to a cgroup with + * no swap usage. Threshold killing only cgroups with more than threshold swap usage. */ + for (int i = 0; i < n; i++) { + /* Skip over cgroups with not enough swap usage. Don't break since there might be "avoid" + * cgroups at the end. */ + if (sorted[i]->swap_usage <= threshold_usage) + continue; + + r = oomd_cgroup_kill(sorted[i]->path, /* recurse= */ true, /* dry_run= */ dry_run); + if (r == -ENOMEM) + return r; /* Treat oom as a hard error */ + if (r < 0) { + if (ret == 0) + ret = r; + continue; /* Try to find something else to kill */ + } + + dump_until = MAX(dump_until, i + 1); + char *selected = strdup(sorted[i]->path); + if (!selected) + return -ENOMEM; + *ret_selected = selected; + ret = r; + break; + } + + dump_kill_candidates(sorted, n, dump_until, oomd_dump_swap_cgroup_context); + + return ret; +} + +int oomd_cgroup_context_acquire(const char *path, OomdCGroupContext **ret) { + _cleanup_(oomd_cgroup_context_freep) OomdCGroupContext *ctx = NULL; + _cleanup_free_ char *p = NULL, *val = NULL; + bool is_root; + int r; + + assert(path); + assert(ret); + + ctx = new0(OomdCGroupContext, 1); + if (!ctx) + return -ENOMEM; + + is_root = empty_or_root(path); + ctx->preference = MANAGED_OOM_PREFERENCE_NONE; + + r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, path, "memory.pressure", &p); + if (r < 0) + return log_debug_errno(r, "Error getting cgroup memory pressure path from %s: %m", path); + + r = read_resource_pressure(p, PRESSURE_TYPE_FULL, &ctx->memory_pressure); + if (r < 0) + return log_debug_errno(r, "Error parsing memory pressure from %s: %m", p); + + if (is_root) { + r = procfs_memory_get_used(&ctx->current_memory_usage); + if (r < 0) + return log_debug_errno(r, "Error getting memory used from procfs: %m"); + } else { + r = cg_get_attribute_as_uint64(SYSTEMD_CGROUP_CONTROLLER, path, "memory.current", &ctx->current_memory_usage); + if (r < 0) + return log_debug_errno(r, "Error getting memory.current from %s: %m", path); + + r = cg_get_attribute_as_uint64(SYSTEMD_CGROUP_CONTROLLER, path, "memory.min", &ctx->memory_min); + if (r < 0) + return log_debug_errno(r, "Error getting memory.min from %s: %m", path); + + r = cg_get_attribute_as_uint64(SYSTEMD_CGROUP_CONTROLLER, path, "memory.low", &ctx->memory_low); + if (r < 0) + return log_debug_errno(r, "Error getting memory.low from %s: %m", path); + + r = cg_get_attribute_as_uint64(SYSTEMD_CGROUP_CONTROLLER, path, "memory.swap.current", &ctx->swap_usage); + if (r == -ENODATA) + /* The kernel can be compiled without support for memory.swap.* files, + * or it can be disabled with boot param 'swapaccount=0' */ + log_once(LOG_WARNING, "No kernel support for memory.swap.current from %s (try boot param swapaccount=1), ignoring.", path); + else if (r < 0) + return log_debug_errno(r, "Error getting memory.swap.current from %s: %m", path); + + r = cg_get_keyed_attribute(SYSTEMD_CGROUP_CONTROLLER, path, "memory.stat", STRV_MAKE("pgscan"), &val); + if (r < 0) + return log_debug_errno(r, "Error getting pgscan from memory.stat under %s: %m", path); + + r = safe_atou64(val, &ctx->pgscan); + if (r < 0) + return log_debug_errno(r, "Error converting pgscan value to uint64_t: %m"); + } + + ctx->path = strdup(empty_to_root(path)); + if (!ctx->path) + return -ENOMEM; + + *ret = TAKE_PTR(ctx); + return 0; +} + +int oomd_system_context_acquire(const char *proc_meminfo_path, OomdSystemContext *ret) { + _cleanup_fclose_ FILE *f = NULL; + unsigned field_filled = 0; + OomdSystemContext ctx = {}; + uint64_t mem_available, swap_free; + int r; + + enum { + MEM_TOTAL = 1U << 0, + MEM_AVAILABLE = 1U << 1, + SWAP_TOTAL = 1U << 2, + SWAP_FREE = 1U << 3, + ALL = MEM_TOTAL|MEM_AVAILABLE|SWAP_TOTAL|SWAP_FREE, + }; + + assert(proc_meminfo_path); + assert(ret); + + f = fopen(proc_meminfo_path, "re"); + if (!f) + return -errno; + + for (;;) { + _cleanup_free_ char *line = NULL; + char *word; + + r = read_line(f, LONG_LINE_MAX, &line); + if (r < 0) + return r; + if (r == 0) + return -EINVAL; + + if ((word = startswith(line, "MemTotal:"))) { + field_filled |= MEM_TOTAL; + r = convert_meminfo_value_to_uint64_bytes(word, &ctx.mem_total); + } else if ((word = startswith(line, "MemAvailable:"))) { + field_filled |= MEM_AVAILABLE; + r = convert_meminfo_value_to_uint64_bytes(word, &mem_available); + } else if ((word = startswith(line, "SwapTotal:"))) { + field_filled |= SWAP_TOTAL; + r = convert_meminfo_value_to_uint64_bytes(word, &ctx.swap_total); + } else if ((word = startswith(line, "SwapFree:"))) { + field_filled |= SWAP_FREE; + r = convert_meminfo_value_to_uint64_bytes(word, &swap_free); + } else + continue; + + if (r < 0) + return log_debug_errno(r, "Error converting '%s' from %s to uint64_t: %m", line, proc_meminfo_path); + + if (field_filled == ALL) + break; + } + + if (field_filled != ALL) + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), "%s is missing expected fields", proc_meminfo_path); + + if (mem_available > ctx.mem_total) + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), + "MemAvailable (%" PRIu64 ") cannot be greater than MemTotal (%" PRIu64 ") %m", + mem_available, + ctx.mem_total); + + if (swap_free > ctx.swap_total) + return log_debug_errno(SYNTHETIC_ERRNO(EINVAL), + "SwapFree (%" PRIu64 ") cannot be greater than SwapTotal (%" PRIu64 ") %m", + swap_free, + ctx.swap_total); + + ctx.mem_used = ctx.mem_total - mem_available; + ctx.swap_used = ctx.swap_total - swap_free; + + *ret = ctx; + return 0; +} + +int oomd_insert_cgroup_context(Hashmap *old_h, Hashmap *new_h, const char *path) { + _cleanup_(oomd_cgroup_context_freep) OomdCGroupContext *curr_ctx = NULL; + OomdCGroupContext *old_ctx; + int r; + + assert(new_h); + assert(path); + + path = empty_to_root(path); + + r = oomd_cgroup_context_acquire(path, &curr_ctx); + if (r < 0) + return log_debug_errno(r, "Failed to get OomdCGroupContext for %s: %m", path); + + assert_se(streq(path, curr_ctx->path)); + + old_ctx = hashmap_get(old_h, path); + if (old_ctx) { + curr_ctx->last_pgscan = old_ctx->pgscan; + curr_ctx->mem_pressure_limit = old_ctx->mem_pressure_limit; + curr_ctx->mem_pressure_limit_hit_start = old_ctx->mem_pressure_limit_hit_start; + curr_ctx->last_had_mem_reclaim = old_ctx->last_had_mem_reclaim; + } + + if (oomd_pgscan_rate(curr_ctx) > 0) + curr_ctx->last_had_mem_reclaim = now(CLOCK_MONOTONIC); + + r = hashmap_put(new_h, curr_ctx->path, curr_ctx); + if (r < 0) + return r; + + TAKE_PTR(curr_ctx); + return 0; +} + +void oomd_update_cgroup_contexts_between_hashmaps(Hashmap *old_h, Hashmap *curr_h) { + OomdCGroupContext *ctx; + + assert(old_h); + assert(curr_h); + + HASHMAP_FOREACH(ctx, curr_h) { + OomdCGroupContext *old_ctx; + + old_ctx = hashmap_get(old_h, ctx->path); + if (!old_ctx) + continue; + + ctx->last_pgscan = old_ctx->pgscan; + ctx->mem_pressure_limit = old_ctx->mem_pressure_limit; + ctx->mem_pressure_limit_hit_start = old_ctx->mem_pressure_limit_hit_start; + ctx->last_had_mem_reclaim = old_ctx->last_had_mem_reclaim; + + if (oomd_pgscan_rate(ctx) > 0) + ctx->last_had_mem_reclaim = now(CLOCK_MONOTONIC); + } +} + +void oomd_dump_swap_cgroup_context(const OomdCGroupContext *ctx, FILE *f, const char *prefix) { + assert(ctx); + assert(f); + + if (!empty_or_root(ctx->path)) + fprintf(f, + "%sPath: %s\n" + "%s\tSwap Usage: %s\n", + strempty(prefix), ctx->path, + strempty(prefix), FORMAT_BYTES(ctx->swap_usage)); + else + fprintf(f, + "%sPath: %s\n" + "%s\tSwap Usage: (see System Context)\n", + strempty(prefix), ctx->path, + strempty(prefix)); +} + +void oomd_dump_memory_pressure_cgroup_context(const OomdCGroupContext *ctx, FILE *f, const char *prefix) { + assert(ctx); + assert(f); + + fprintf(f, + "%sPath: %s\n" + "%s\tMemory Pressure Limit: %lu.%02lu%%\n" + "%s\tPressure: Avg10: %lu.%02lu Avg60: %lu.%02lu Avg300: %lu.%02lu Total: %s\n" + "%s\tCurrent Memory Usage: %s\n", + strempty(prefix), ctx->path, + strempty(prefix), LOADAVG_INT_SIDE(ctx->mem_pressure_limit), LOADAVG_DECIMAL_SIDE(ctx->mem_pressure_limit), + strempty(prefix), + LOADAVG_INT_SIDE(ctx->memory_pressure.avg10), LOADAVG_DECIMAL_SIDE(ctx->memory_pressure.avg10), + LOADAVG_INT_SIDE(ctx->memory_pressure.avg60), LOADAVG_DECIMAL_SIDE(ctx->memory_pressure.avg60), + LOADAVG_INT_SIDE(ctx->memory_pressure.avg300), LOADAVG_DECIMAL_SIDE(ctx->memory_pressure.avg300), + FORMAT_TIMESPAN(ctx->memory_pressure.total, USEC_PER_SEC), + strempty(prefix), FORMAT_BYTES(ctx->current_memory_usage)); + + if (!empty_or_root(ctx->path)) + fprintf(f, + "%s\tMemory Min: %s\n" + "%s\tMemory Low: %s\n" + "%s\tPgscan: %" PRIu64 "\n" + "%s\tLast Pgscan: %" PRIu64 "\n", + strempty(prefix), FORMAT_BYTES_CGROUP_PROTECTION(ctx->memory_min), + strempty(prefix), FORMAT_BYTES_CGROUP_PROTECTION(ctx->memory_low), + strempty(prefix), ctx->pgscan, + strempty(prefix), ctx->last_pgscan); +} + +void oomd_dump_system_context(const OomdSystemContext *ctx, FILE *f, const char *prefix) { + assert(ctx); + assert(f); + + fprintf(f, + "%sMemory: Used: %s Total: %s\n" + "%sSwap: Used: %s Total: %s\n", + strempty(prefix), + FORMAT_BYTES(ctx->mem_used), + FORMAT_BYTES(ctx->mem_total), + strempty(prefix), + FORMAT_BYTES(ctx->swap_used), + FORMAT_BYTES(ctx->swap_total)); +} |