From a285909f471d6703a04b2b3942c352e27131c92b Mon Sep 17 00:00:00 2001 From: Hyeonggon Yoo <42.hyeyoo@gmail.com> Date: Wed, 6 Apr 2022 15:00:03 +0900 Subject: mm/slub, kunit: Make slub_kunit unaffected by user specified flags slub_kunit does not expect other debugging flags to be set when running tests. When SLAB_RED_ZONE flag is set globally, test fails because the flag affects number of errors reported. To make slub_kunit unaffected by user specified debugging flags, introduce SLAB_NO_USER_FLAGS to ignore them. With this flag, only flags specified in the code are used and others are ignored. Suggested-by: Vlastimil Babka Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Signed-off-by: Vlastimil Babka Link: https://lore.kernel.org/r/Yk0sY9yoJhFEXWOg@hyeyoo --- lib/slub_kunit.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'lib') diff --git a/lib/slub_kunit.c b/lib/slub_kunit.c index 8662dc6cb509..7a0564d7cb7a 100644 --- a/lib/slub_kunit.c +++ b/lib/slub_kunit.c @@ -12,7 +12,7 @@ static int slab_errors; static void test_clobber_zone(struct kunit *test) { struct kmem_cache *s = kmem_cache_create("TestSlub_RZ_alloc", 64, 0, - SLAB_RED_ZONE, NULL); + SLAB_RED_ZONE|SLAB_NO_USER_FLAGS, NULL); u8 *p = kmem_cache_alloc(s, GFP_KERNEL); kasan_disable_current(); @@ -30,7 +30,7 @@ static void test_clobber_zone(struct kunit *test) static void test_next_pointer(struct kunit *test) { struct kmem_cache *s = kmem_cache_create("TestSlub_next_ptr_free", 64, 0, - SLAB_POISON, NULL); + SLAB_POISON|SLAB_NO_USER_FLAGS, NULL); u8 *p = kmem_cache_alloc(s, GFP_KERNEL); unsigned long tmp; unsigned long *ptr_addr; @@ -75,7 +75,7 @@ static void test_next_pointer(struct kunit *test) static void test_first_word(struct kunit *test) { struct kmem_cache *s = kmem_cache_create("TestSlub_1th_word_free", 64, 0, - SLAB_POISON, NULL); + SLAB_POISON|SLAB_NO_USER_FLAGS, NULL); u8 *p = kmem_cache_alloc(s, GFP_KERNEL); kmem_cache_free(s, p); @@ -90,7 +90,7 @@ static void test_first_word(struct kunit *test) static void test_clobber_50th_byte(struct kunit *test) { struct kmem_cache *s = kmem_cache_create("TestSlub_50th_word_free", 64, 0, - SLAB_POISON, NULL); + SLAB_POISON|SLAB_NO_USER_FLAGS, NULL); u8 *p = kmem_cache_alloc(s, GFP_KERNEL); kmem_cache_free(s, p); @@ -106,7 +106,7 @@ static void test_clobber_50th_byte(struct kunit *test) static void test_clobber_redzone_free(struct kunit *test) { struct kmem_cache *s = kmem_cache_create("TestSlub_RZ_free", 64, 0, - SLAB_RED_ZONE, NULL); + SLAB_RED_ZONE|SLAB_NO_USER_FLAGS, NULL); u8 *p = kmem_cache_alloc(s, GFP_KERNEL); kasan_disable_current(); -- cgit v1.2.3 From a5f1783be29adae15666fd803efd7d2979130869 Mon Sep 17 00:00:00 2001 From: Vlastimil Babka Date: Wed, 2 Mar 2022 12:02:22 +0100 Subject: lib/stackdepot: allow requesting early initialization dynamically In a later patch we want to add stackdepot support for object owner tracking in slub caches, which is enabled by slub_debug boot parameter. This creates a bootstrap problem as some caches are created early in boot when slab_is_available() is false and thus stack_depot_init() tries to use memblock. But, as reported by Hyeonggon Yoo [1] we are already beyond memblock_free_all(). Ideally memblock allocation should fail, yet it succeeds, but later the system crashes, which is a separately handled issue. To resolve this boostrap issue in a robust way, this patch adds another way to request stack_depot_early_init(), which happens at a well-defined point of time. In addition to build-time CONFIG_STACKDEPOT_ALWAYS_INIT, code that's e.g. processing boot parameters (which happens early enough) can call a new function stack_depot_want_early_init(), which sets a flag that stack_depot_early_init() will check. In this patch we also convert page_owner to this approach. While it doesn't have the bootstrap issue as slub, it's also a functionality enabled by a boot param and can thus request stack_depot_early_init() with memblock allocation instead of later initialization with kvmalloc(). As suggested by Mike, make stack_depot_early_init() only attempt memblock allocation and stack_depot_init() only attempt kvmalloc(). Also change the latter to kvcalloc(). In both cases we can lose the explicit array zeroing, which the allocations do already. As suggested by Marco, provide empty implementations of the init functions for !CONFIG_STACKDEPOT builds to simplify the callers. [1] https://lore.kernel.org/all/YhnUcqyeMgCrWZbd@ip-172-31-19-208.ap-northeast-1.compute.internal/ Reported-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Suggested-by: Mike Rapoport Suggested-by: Marco Elver Signed-off-by: Vlastimil Babka Reviewed-by: Marco Elver Reviewed-and-tested-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Reviewed-by: Mike Rapoport Acked-by: David Rientjes --- include/linux/stackdepot.h | 26 +++++++++++++++--- lib/stackdepot.c | 67 +++++++++++++++++++++++++++++++--------------- mm/page_owner.c | 9 ++++--- 3 files changed, 73 insertions(+), 29 deletions(-) (limited to 'lib') diff --git a/include/linux/stackdepot.h b/include/linux/stackdepot.h index 17f992fe6355..bc2797955de9 100644 --- a/include/linux/stackdepot.h +++ b/include/linux/stackdepot.h @@ -20,18 +20,36 @@ depot_stack_handle_t __stack_depot_save(unsigned long *entries, gfp_t gfp_flags, bool can_alloc); /* - * Every user of stack depot has to call this during its own init when it's - * decided that it will be calling stack_depot_save() later. + * Every user of stack depot has to call stack_depot_init() during its own init + * when it's decided that it will be calling stack_depot_save() later. This is + * recommended for e.g. modules initialized later in the boot process, when + * slab_is_available() is true. * * The alternative is to select STACKDEPOT_ALWAYS_INIT to have stack depot * enabled as part of mm_init(), for subsystems where it's known at compile time * that stack depot will be used. + * + * Another alternative is to call stack_depot_want_early_init(), when the + * decision to use stack depot is taken e.g. when evaluating kernel boot + * parameters, which precedes the enablement point in mm_init(). + * + * stack_depot_init() and stack_depot_want_early_init() can be called regardless + * of CONFIG_STACKDEPOT and are no-op when disabled. The actual save/fetch/print + * functions should only be called from code that makes sure CONFIG_STACKDEPOT + * is enabled. */ +#ifdef CONFIG_STACKDEPOT int stack_depot_init(void); -#ifdef CONFIG_STACKDEPOT_ALWAYS_INIT -static inline int stack_depot_early_init(void) { return stack_depot_init(); } +void __init stack_depot_want_early_init(void); + +/* This is supposed to be called only from mm_init() */ +int __init stack_depot_early_init(void); #else +static inline int stack_depot_init(void) { return 0; } + +static inline void stack_depot_want_early_init(void) { } + static inline int stack_depot_early_init(void) { return 0; } #endif diff --git a/lib/stackdepot.c b/lib/stackdepot.c index bf5ba9af0500..5ca0d086ef4a 100644 --- a/lib/stackdepot.c +++ b/lib/stackdepot.c @@ -66,6 +66,9 @@ struct stack_record { unsigned long entries[]; /* Variable-sized array of entries. */ }; +static bool __stack_depot_want_early_init __initdata = IS_ENABLED(CONFIG_STACKDEPOT_ALWAYS_INIT); +static bool __stack_depot_early_init_passed __initdata; + static void *stack_slabs[STACK_ALLOC_MAX_SLABS]; static int depot_index; @@ -162,38 +165,58 @@ static int __init is_stack_depot_disabled(char *str) } early_param("stack_depot_disable", is_stack_depot_disabled); -/* - * __ref because of memblock_alloc(), which will not be actually called after - * the __init code is gone, because at that point slab_is_available() is true - */ -__ref int stack_depot_init(void) +void __init stack_depot_want_early_init(void) +{ + /* Too late to request early init now */ + WARN_ON(__stack_depot_early_init_passed); + + __stack_depot_want_early_init = true; +} + +int __init stack_depot_early_init(void) +{ + size_t size; + + /* This is supposed to be called only once, from mm_init() */ + if (WARN_ON(__stack_depot_early_init_passed)) + return 0; + + __stack_depot_early_init_passed = true; + + if (!__stack_depot_want_early_init || stack_depot_disable) + return 0; + + size = (STACK_HASH_SIZE * sizeof(struct stack_record *)); + pr_info("Stack Depot early init allocating hash table with memblock_alloc, %zu bytes\n", + size); + stack_table = memblock_alloc(size, SMP_CACHE_BYTES); + + if (!stack_table) { + pr_err("Stack Depot hash table allocation failed, disabling\n"); + stack_depot_disable = true; + return -ENOMEM; + } + + return 0; +} + +int stack_depot_init(void) { static DEFINE_MUTEX(stack_depot_init_mutex); + int ret = 0; mutex_lock(&stack_depot_init_mutex); if (!stack_depot_disable && !stack_table) { - size_t size = (STACK_HASH_SIZE * sizeof(struct stack_record *)); - int i; - - if (slab_is_available()) { - pr_info("Stack Depot allocating hash table with kvmalloc\n"); - stack_table = kvmalloc(size, GFP_KERNEL); - } else { - pr_info("Stack Depot allocating hash table with memblock_alloc\n"); - stack_table = memblock_alloc(size, SMP_CACHE_BYTES); - } - if (stack_table) { - for (i = 0; i < STACK_HASH_SIZE; i++) - stack_table[i] = NULL; - } else { + pr_info("Stack Depot allocating hash table with kvcalloc\n"); + stack_table = kvcalloc(STACK_HASH_SIZE, sizeof(struct stack_record *), GFP_KERNEL); + if (!stack_table) { pr_err("Stack Depot hash table allocation failed, disabling\n"); stack_depot_disable = true; - mutex_unlock(&stack_depot_init_mutex); - return -ENOMEM; + ret = -ENOMEM; } } mutex_unlock(&stack_depot_init_mutex); - return 0; + return ret; } EXPORT_SYMBOL_GPL(stack_depot_init); diff --git a/mm/page_owner.c b/mm/page_owner.c index fb3a05fdebdb..2743062e92c2 100644 --- a/mm/page_owner.c +++ b/mm/page_owner.c @@ -45,7 +45,12 @@ static void init_early_allocated_pages(void); static int __init early_page_owner_param(char *buf) { - return kstrtobool(buf, &page_owner_enabled); + int ret = kstrtobool(buf, &page_owner_enabled); + + if (page_owner_enabled) + stack_depot_want_early_init(); + + return ret; } early_param("page_owner", early_page_owner_param); @@ -83,8 +88,6 @@ static __init void init_page_owner(void) if (!page_owner_enabled) return; - stack_depot_init(); - register_dummy_stack(); register_failure_stack(); register_early_stack(); -- cgit v1.2.3 From 5cf909c553e9efed573811de4b3f5172898d5515 Mon Sep 17 00:00:00 2001 From: Oliver Glitta Date: Wed, 7 Jul 2021 18:07:47 -0700 Subject: mm/slub: use stackdepot to save stack trace in objects Many stack traces are similar so there are many similar arrays. Stackdepot saves each unique stack only once. Replace field addrs in struct track with depot_stack_handle_t handle. Use stackdepot to save stack trace. The benefits are smaller memory overhead and possibility to aggregate per-cache statistics in the following patch using the stackdepot handle instead of matching stacks manually. [ vbabka@suse.cz: rebase to 5.17-rc1 and adjust accordingly ] This was initially merged as commit 788691464c29 and reverted by commit ae14c63a9f20 due to several issues, that should now be fixed. The problem of unconditional memory overhead by stackdepot has been addressed by commit 2dba5eb1c73b ("lib/stackdepot: allow optional init and stack_table allocation by kvmalloc()"), so the dependency on stackdepot will result in extra memory usage only when a slab cache tracking is actually enabled, and not for all CONFIG_SLUB_DEBUG builds. The build failures on some architectures were also addressed, and the reported issue with xfs/433 test did not reproduce on 5.17-rc1 with this patch. Signed-off-by: Oliver Glitta Signed-off-by: Vlastimil Babka Reviewed-and-tested-by: Hyeonggon Yoo <42.hyeyoo@gmail.com> Acked-by: David Rientjes Cc: David Rientjes Cc: Christoph Lameter Cc: Pekka Enberg Cc: Joonsoo Kim --- init/Kconfig | 1 + lib/Kconfig.debug | 1 + mm/slab_common.c | 5 ++++ mm/slub.c | 71 +++++++++++++++++++++++++++++++------------------------ 4 files changed, 47 insertions(+), 31 deletions(-) (limited to 'lib') diff --git a/init/Kconfig b/init/Kconfig index ddcbefe535e9..adc57f989d87 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -1875,6 +1875,7 @@ config SLUB_DEBUG default y bool "Enable SLUB debugging support" if EXPERT depends on SLUB && SYSFS + select STACKDEPOT if STACKTRACE_SUPPORT help SLUB has extensive debug support features. Disabling these can result in significant savings in code size. This also disables diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index 075cd25363ac..78d6139111cd 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -709,6 +709,7 @@ config DEBUG_SLAB config SLUB_DEBUG_ON bool "SLUB debugging on by default" depends on SLUB && SLUB_DEBUG + select STACKDEPOT_ALWAYS_INIT if STACKTRACE_SUPPORT default n help Boot with debugging on by default. SLUB boots by default with diff --git a/mm/slab_common.c b/mm/slab_common.c index 6ee64d6208b3..73943479a2b7 100644 --- a/mm/slab_common.c +++ b/mm/slab_common.c @@ -24,6 +24,7 @@ #include #include #include +#include #define CREATE_TRACE_POINTS #include @@ -314,9 +315,13 @@ kmem_cache_create_usercopy(const char *name, * If no slub_debug was enabled globally, the static key is not yet * enabled by setup_slub_debug(). Enable it if the cache is being * created with any of the debugging flags passed explicitly. + * It's also possible that this is the first cache created with + * SLAB_STORE_USER and we should init stack_depot for it. */ if (flags & SLAB_DEBUG_FLAGS) static_branch_enable(&slub_debug_enabled); + if (flags & SLAB_STORE_USER) + stack_depot_init(); #endif mutex_lock(&slab_mutex); diff --git a/mm/slub.c b/mm/slub.c index cd4fd0159911..98c1450c23f0 100644 --- a/mm/slub.c +++ b/mm/slub.c @@ -26,6 +26,7 @@ #include #include #include +#include #include #include #include @@ -264,8 +265,8 @@ static inline bool kmem_cache_has_cpu_partial(struct kmem_cache *s) #define TRACK_ADDRS_COUNT 16 struct track { unsigned long addr; /* Called from address */ -#ifdef CONFIG_STACKTRACE - unsigned long addrs[TRACK_ADDRS_COUNT]; /* Called from address */ +#ifdef CONFIG_STACKDEPOT + depot_stack_handle_t handle; #endif int cpu; /* Was running on cpu */ int pid; /* Pid context */ @@ -724,22 +725,19 @@ static struct track *get_track(struct kmem_cache *s, void *object, return kasan_reset_tag(p + alloc); } -static void set_track(struct kmem_cache *s, void *object, +static void noinline set_track(struct kmem_cache *s, void *object, enum track_item alloc, unsigned long addr) { struct track *p = get_track(s, object, alloc); -#ifdef CONFIG_STACKTRACE +#ifdef CONFIG_STACKDEPOT + unsigned long entries[TRACK_ADDRS_COUNT]; unsigned int nr_entries; - metadata_access_enable(); - nr_entries = stack_trace_save(kasan_reset_tag(p->addrs), - TRACK_ADDRS_COUNT, 3); - metadata_access_disable(); - - if (nr_entries < TRACK_ADDRS_COUNT) - p->addrs[nr_entries] = 0; + nr_entries = stack_trace_save(entries, ARRAY_SIZE(entries), 3); + p->handle = stack_depot_save(entries, nr_entries, GFP_NOWAIT); #endif + p->addr = addr; p->cpu = smp_processor_id(); p->pid = current->pid; @@ -759,20 +757,19 @@ static void init_tracking(struct kmem_cache *s, void *object) static void print_track(const char *s, struct track *t, unsigned long pr_time) { + depot_stack_handle_t handle __maybe_unused; + if (!t->addr) return; pr_err("%s in %pS age=%lu cpu=%u pid=%d\n", s, (void *)t->addr, pr_time - t->when, t->cpu, t->pid); -#ifdef CONFIG_STACKTRACE - { - int i; - for (i = 0; i < TRACK_ADDRS_COUNT; i++) - if (t->addrs[i]) - pr_err("\t%pS\n", (void *)t->addrs[i]); - else - break; - } +#ifdef CONFIG_STACKDEPOT + handle = READ_ONCE(t->handle); + if (handle) + stack_depot_print(handle); + else + pr_err("object allocation/free stack trace missing\n"); #endif } @@ -1532,6 +1529,8 @@ static int __init setup_slub_debug(char *str) global_slub_debug_changed = true; } else { slab_list_specified = true; + if (flags & SLAB_STORE_USER) + stack_depot_want_early_init(); } } @@ -1549,6 +1548,8 @@ static int __init setup_slub_debug(char *str) } out: slub_debug = global_flags; + if (slub_debug & SLAB_STORE_USER) + stack_depot_want_early_init(); if (slub_debug != 0 || slub_debug_string) static_branch_enable(&slub_debug_enabled); else @@ -4342,18 +4343,26 @@ void kmem_obj_info(struct kmem_obj_info *kpp, void *object, struct slab *slab) objp = fixup_red_left(s, objp); trackp = get_track(s, objp, TRACK_ALLOC); kpp->kp_ret = (void *)trackp->addr; -#ifdef CONFIG_STACKTRACE - for (i = 0; i < KS_ADDRS_COUNT && i < TRACK_ADDRS_COUNT; i++) { - kpp->kp_stack[i] = (void *)trackp->addrs[i]; - if (!kpp->kp_stack[i]) - break; - } +#ifdef CONFIG_STACKDEPOT + { + depot_stack_handle_t handle; + unsigned long *entries; + unsigned int nr_entries; + + handle = READ_ONCE(trackp->handle); + if (handle) { + nr_entries = stack_depot_fetch(handle, &entries); + for (i = 0; i < KS_ADDRS_COUNT && i < nr_entries; i++) + kpp->kp_stack[i] = (void *)entries[i]; + } - trackp = get_track(s, objp, TRACK_FREE); - for (i = 0; i < KS_ADDRS_COUNT && i < TRACK_ADDRS_COUNT; i++) { - kpp->kp_free_stack[i] = (void *)trackp->addrs[i]; - if (!kpp->kp_free_stack[i]) - break; + trackp = get_track(s, objp, TRACK_FREE); + handle = READ_ONCE(trackp->handle); + if (handle) { + nr_entries = stack_depot_fetch(handle, &entries); + for (i = 0; i < KS_ADDRS_COUNT && i < nr_entries; i++) + kpp->kp_free_stack[i] = (void *)entries[i]; + } } #endif #endif -- cgit v1.2.3