2006-08-12 21:59:44

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

This is a new version (0.9) of the kernel memory leak detector. See
the Documentation/kmemleak.txt file for a more detailed
description. The patches are downloadable from (the whole patch or the
broken-out series):

http://homepage.ntlworld.com/cmarinas/kmemleak/patch-2.6.18-rc4-kmemleak-0.9.bz2
http://homepage.ntlworld.com/cmarinas/kmemleak/broken-out/patches-kmemleak-0.9.tar.bz2

What's new in this version:

- updated for 2.6.18-rc4
- reviewers comments implemented
- the number of reports are limited to avoid soft-lockups in case of a
serious memory leak
- some bug-fixes (including one that was preventing the page-aligned
blocks from being reported)

To do:

- testing on a wider range of platforms and configurations
- support for ioremap tracking (once the generic ioremap patches are
merged)
- eliminate the task stacks scanning (if possible, by marking the
allocated blocks as temporary until the return to user-space -
Ingo's suggestion)
- precise type identification (after first assessing the efficiency of
the current method as it requires changes to the kernel API)

--
Catalin


2006-08-12 22:00:41

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 05/10] Add kmemleak support for i386

From: Catalin Marinas <[email protected]>

This patch modifies the vmlinux.lds.S script and adds the backtrace support
for i386 to be used with kmemleak.

Signed-off-by: Catalin Marinas <[email protected]>
---

arch/i386/kernel/vmlinux.lds.S | 4 ++++
include/asm-i386/processor.h | 12 ++++++++++++
include/asm-i386/thread_info.h | 10 +++++++++-
3 files changed, 25 insertions(+), 1 deletions(-)

diff --git a/arch/i386/kernel/vmlinux.lds.S b/arch/i386/kernel/vmlinux.lds.S
index 2d4f138..34fccf1 100644
--- a/arch/i386/kernel/vmlinux.lds.S
+++ b/arch/i386/kernel/vmlinux.lds.S
@@ -45,6 +45,7 @@ SECTIONS
__tracedata_end = .;

/* writeable */
+ _sdata = .; /* Start of data section */
.data : AT(ADDR(.data) - LOAD_OFFSET) { /* Data */
*(.data)
CONSTRUCTORS
@@ -156,6 +157,9 @@ #endif
__per_cpu_start = .;
.data.percpu : AT(ADDR(.data.percpu) - LOAD_OFFSET) { *(.data.percpu) }
__per_cpu_end = .;
+ __memleak_offsets_start = .;
+ .init.memleak_offsets : AT(ADDR(.init.memleak_offsets) - LOAD_OFFSET) { *(.init.memleak_offsets) }
+ __memleak_offsets_end = .;
. = ALIGN(4096);
__init_end = .;
/* freed after init ends here */
diff --git a/include/asm-i386/processor.h b/include/asm-i386/processor.h
index b32346d..29626e3 100644
--- a/include/asm-i386/processor.h
+++ b/include/asm-i386/processor.h
@@ -731,4 +731,16 @@ extern unsigned long boot_option_idle_ov
extern void enable_sep_cpu(void);
extern int sysenter_setup(void);

+#ifdef CONFIG_FRAME_POINTER
+static inline unsigned long arch_call_address(void *frame)
+{
+ return *(unsigned long *)(frame + 4);
+}
+
+static inline void *arch_prev_frame(void *frame)
+{
+ return *(void **)frame;
+}
+#endif
+
#endif /* __ASM_I386_PROCESSOR_H */
diff --git a/include/asm-i386/thread_info.h b/include/asm-i386/thread_info.h
index 54d6d7a..054553f 100644
--- a/include/asm-i386/thread_info.h
+++ b/include/asm-i386/thread_info.h
@@ -100,12 +100,20 @@ #define alloc_thread_info(tsk) \
struct thread_info *ret; \
\
ret = kmalloc(THREAD_SIZE, GFP_KERNEL); \
+ memleak_ignore(ret); \
if (ret) \
memset(ret, 0, THREAD_SIZE); \
ret; \
})
#else
-#define alloc_thread_info(tsk) kmalloc(THREAD_SIZE, GFP_KERNEL)
+#define alloc_thread_info(tsk) \
+ ({ \
+ struct thread_info *ret; \
+ \
+ ret = kmalloc(THREAD_SIZE, GFP_KERNEL); \
+ memleak_ignore(ret); \
+ ret; \
+ })
#endif

#define free_thread_info(info) kfree(info)

2006-08-12 22:00:00

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 01/10] Base support for kmemleak

From: Catalin Marinas <[email protected]>

This patch adds the base support for the kernel memory leak detector. It
traces the memory allocation/freeing in a way similar to the Boehm's
conservative garbage collector, the difference being that the orphan
pointers are not freed but only shown in /proc/memleak. Enabling this
feature would introduce an overhead to memory allocations.

Signed-off-by: Catalin Marinas <[email protected]>
---

include/linux/kernel.h | 7
include/linux/memleak.h | 111 ++++
init/main.c | 3
lib/Kconfig.debug | 80 +++
mm/Makefile | 2
mm/memleak.c | 1320 +++++++++++++++++++++++++++++++++++++++++++++++
6 files changed, 1520 insertions(+), 3 deletions(-)

diff --git a/include/linux/kernel.h b/include/linux/kernel.h
index 181c69c..d071716 100644
--- a/include/linux/kernel.h
+++ b/include/linux/kernel.h
@@ -13,6 +13,7 @@ #include <linux/stddef.h>
#include <linux/types.h>
#include <linux/compiler.h>
#include <linux/bitops.h>
+#include <linux/memleak.h>
#include <asm/byteorder.h>
#include <asm/bug.h>

@@ -291,9 +292,13 @@ #define max_t(type,x,y) \
* @member: the name of the member within the struct.
*
*/
-#define container_of(ptr, type, member) ({ \
+#define __container_of(ptr, type, member) ({ \
const typeof( ((type *)0)->member ) *__mptr = (ptr); \
(type *)( (char *)__mptr - offsetof(type,member) );})
+#define container_of(ptr, type, member) ({ \
+ DECLARE_MEMLEAK_OFFSET(container_of, type, member); \
+ __container_of(ptr, type, member); \
+})

/*
* Check at compile time that something is of a particular type.
diff --git a/include/linux/memleak.h b/include/linux/memleak.h
new file mode 100644
index 0000000..39669bf
--- /dev/null
+++ b/include/linux/memleak.h
@@ -0,0 +1,111 @@
+/*
+ * include/linux/memleak.h
+ *
+ * Copyright (C) 2006 ARM Limited
+ * Written by Catalin Marinas <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef __MEMLEAK_H
+#define __MEMLEAK_H
+
+#include <linux/stddef.h>
+
+struct memleak_offset {
+ unsigned long type_id;
+ unsigned long member_type_id;
+ unsigned long offset;
+};
+
+/* type id approximation */
+#define ml_guess_typeid(size) ((unsigned long)(size))
+#define ml_typeid(type) ml_guess_typeid(sizeof(type))
+#define ml_sizeof(typeid) ((size_t)(typeid))
+
+#ifdef CONFIG_DEBUG_MEMLEAK
+
+/* if offsetof(type, member) is not a constant known at compile time,
+ * just use 0 instead since we cannot add it to the
+ * .init.memleak_offsets section
+ */
+#define memleak_offsetof(type, member) \
+ (__builtin_constant_p(offsetof(type, member)) ? \
+ offsetof(type, member) : 0)
+
+#define DECLARE_MEMLEAK_OFFSET(name, type, member) \
+ static const struct memleak_offset \
+ __attribute__ ((__section__ (".init.memleak_offsets"))) \
+ __attribute_used__ __memleak_offset__##name = { \
+ ml_typeid(type), \
+ ml_typeid(typeof(((type *)0)->member)), \
+ memleak_offsetof(type, member) \
+ }
+
+extern void memleak_init(void);
+extern void memleak_alloc(const void *ptr, size_t size, int ref_count);
+extern void memleak_free(const void *ptr);
+extern void memleak_padding(const void *ptr, unsigned long offset, size_t size);
+extern void memleak_not_leak(const void *ptr);
+extern void memleak_ignore(const void *ptr);
+extern void memleak_scan_area(const void *ptr, unsigned long offset, size_t length);
+extern void memleak_insert_aliases(struct memleak_offset *ml_off_start,
+ struct memleak_offset *ml_off_end);
+
+static inline void memleak_erase(void **ptr)
+{
+ *ptr = NULL;
+}
+
+#define memleak_container(type, member) { \
+ DECLARE_MEMLEAK_OFFSET(container_of, type, member); \
+}
+
+extern void memleak_typeid_raw(const void *ptr, unsigned long type_id);
+#define memleak_typeid(ptr, type) \
+ memleak_typeid_raw(ptr, ml_typeid(type))
+
+#else
+
+#define DECLARE_MEMLEAK_OFFSET(name, type, member)
+
+static inline void memleak_init(void)
+{ }
+static inline void memleak_alloc(const void *ptr, size_t size, int ref_count)
+{ }
+static inline void memleak_free(const void *ptr)
+{ }
+static inline void memleak_padding(const void *ptr, unsigned long offset, size_t size)
+{ }
+static inline void memleak_not_leak(const void *ptr)
+{ }
+static inline void memleak_ignore(const void *ptr)
+{ }
+static inline void memleak_scan_area(const void *ptr, unsigned long offset, size_t length)
+{ }
+static inline void memleak_insert_aliases(struct memleak_offset *ml_off_start,
+ struct memleak_offset *ml_off_end)
+{ }
+static inline void memleak_erase(void **ptr)
+{ }
+
+#define memleak_container(type, member)
+
+static inline void memleak_typeid_raw(const void *ptr, unsigned long type_id)
+{ }
+#define memleak_typeid(ptr, type)
+
+#endif /* CONFIG_DEBUG_MEMLEAK */
+
+#endif /* __MEMLEAK_H */
diff --git a/init/main.c b/init/main.c
index 8651a72..6497f61 100644
--- a/init/main.c
+++ b/init/main.c
@@ -548,6 +548,8 @@ #endif
cpuset_init_early();
mem_init();
kmem_cache_init();
+ radix_tree_init();
+ memleak_init();
setup_per_cpu_pageset();
numa_policy_init();
if (late_time_init)
@@ -568,7 +570,6 @@ #endif
key_init();
security_init();
vfs_caches_init(num_physpages);
- radix_tree_init();
signals_init();
/* rootfs populating might need page-writeback */
page_writeback_init();
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 554ee68..77ddb7d 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -105,6 +105,86 @@ config DEBUG_SLAB_LEAK
bool "Memory leak debugging"
depends on DEBUG_SLAB

+menuconfig DEBUG_MEMLEAK
+ bool "Kernel memory leak detector"
+ default n
+ depends on EXPERIMENTAL && DEBUG_SLAB
+ select DEBUG_FS
+ help
+ Say Y here if you want to enable the memory leak
+ detector. The memory allocation/freeing is traced in a way
+ similar to the Boehm's conservative garbage collector, the
+ difference being that the orphan pointers are not freed but
+ only shown in /sys/kernel/debug/memleak. Enabling this
+ feature will introduce an overhead to memory
+ allocations. See Documentation/kmemleak.txt for more
+ details.
+
+ In order to access the memleak file, debugfs needs to be
+ mounted (usually at /sys/kernel/debug).
+
+config DEBUG_MEMLEAK_TRACE_LENGTH
+ int "KMemLeak stack trace length"
+ default 4
+ depends on DEBUG_MEMLEAK && FRAME_POINTER
+ help
+ This option sets the length of the stack trace for the
+ allocated pointers tracked by kmemleak.
+
+config DEBUG_MEMLEAK_PREINIT_POINTERS
+ int "KMemLeak pre-init actions buffer size"
+ default 512
+ depends on DEBUG_MEMLEAK
+ help
+ This is the buffer for storing the memory allocation/freeing
+ calls before kmemleak is fully initialized. Each element in
+ the buffer takes 24 bytes on a 32 bit architecture. This
+ buffer will be freed once the system initialization is
+ completed.
+
+config DEBUG_MEMLEAK_SECONDARY_ALIASES
+ bool "Create secondary level pointer aliases"
+ default y
+ depends on DEBUG_MEMLEAK
+ help
+ This option creates aliases for container_of(container_of(member))
+ access to pointers. Disabling this option reduces the chances of
+ false negatives but it can slightly increase the number of false
+ positives.
+
+config DEBUG_MEMLEAK_TASK_STACKS
+ bool "Scan task kernel stacks"
+ default y
+ depends on DEBUG_MEMLEAK
+ help
+ This option enables the scanning of the task kernel
+ stacks. This option can introduce false negatives because of
+ the randomness of stacks content.
+
+ If unsure, say Y.
+
+config DEBUG_MEMLEAK_ORPHAN_FREEING
+ bool "Notify when freeing orphan pointers"
+ default n
+ depends on DEBUG_MEMLEAK
+ help
+ This option enables the notification when pointers
+ considered leaks are freed. The stack dump and the pointer
+ information displayed allow an easier identification of
+ false positives. Use this mainly for debugging kmemleak.
+
+ If unsure, say N.
+
+config DEBUG_MEMLEAK_REPORTS_NR
+ int "Maximum number of reported leaks"
+ default 100
+ depends on DEBUG_MEMLEAK
+ help
+ This option sets the maximum number of leaks reported. If
+ this number is too big and there are leaks to be reported,
+ reading the /sys/kernel/debug/memleak file could lead to
+ some soft-locks.
+
config DEBUG_PREEMPT
bool "Debug preemptible kernel"
depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
diff --git a/mm/Makefile b/mm/Makefile
index 9dd824c..edccbc0 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -23,4 +23,4 @@ obj-$(CONFIG_SLAB) += slab.o
obj-$(CONFIG_MEMORY_HOTPLUG) += memory_hotplug.o
obj-$(CONFIG_FS_XIP) += filemap_xip.o
obj-$(CONFIG_MIGRATION) += migrate.o
-
+obj-$(CONFIG_DEBUG_MEMLEAK) += memleak.o
diff --git a/mm/memleak.c b/mm/memleak.c
new file mode 100644
index 0000000..514280f
--- /dev/null
+++ b/mm/memleak.c
@@ -0,0 +1,1320 @@
+/*
+ * mm/memleak.c
+ *
+ * Copyright (C) 2006 ARM Limited
+ * Written by Catalin Marinas <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+/* #define DEBUG */
+
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/radix-tree.h>
+#include <linux/gfp.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/kallsyms.h>
+#include <linux/mman.h>
+#include <linux/nodemask.h>
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+#include <linux/vmalloc.h>
+#include <linux/mutex.h>
+#include <linux/cpumask.h>
+
+#include <asm/bitops.h>
+#include <asm/sections.h>
+#include <asm/percpu.h>
+#include <asm/processor.h>
+#include <asm/thread_info.h>
+#include <asm/atomic.h>
+
+#include <linux/memleak.h>
+
+#ifdef CONFIG_FRAME_POINTER
+#define MAX_TRACE CONFIG_DEBUG_MEMLEAK_TRACE_LENGTH
+#else
+#define MAX_TRACE 1
+#endif
+
+#define SCAN_BLOCK_SIZE 4096 /* maximum scan length with interrupts disabled */
+#define PREINIT_POINTERS CONFIG_DEBUG_MEMLEAK_PREINIT_POINTERS
+#define BYTES_PER_WORD sizeof(void *)
+
+extern struct memleak_offset __memleak_offsets_start[];
+extern struct memleak_offset __memleak_offsets_end[];
+
+struct memleak_alias {
+ struct hlist_node node;
+ unsigned long offset;
+};
+
+struct memleak_scan_area {
+ struct hlist_node node;
+ unsigned long offset;
+ size_t length;
+};
+
+struct memleak_pointer {
+ unsigned long flags;
+ struct list_head pointer_list;
+ struct list_head gray_list;
+ int use_count;
+ unsigned long pointer;
+ unsigned long offset; /* padding */
+ size_t size;
+ unsigned long type_id;
+ int ref_count; /* the minimum encounters of the value */
+ int count; /* the ecounters of the value */
+ struct hlist_head *alias_list;
+ struct hlist_head area_list; /* areas to be scanned (or empty for all) */
+ unsigned long trace[MAX_TRACE];
+};
+
+enum memleak_action {
+ MEMLEAK_ALLOC,
+ MEMLEAK_FREE,
+ MEMLEAK_PADDING,
+ MEMLEAK_NOT_LEAK,
+ MEMLEAK_IGNORE,
+ MEMLEAK_SCAN_AREA,
+ MEMLEAK_TYPEID
+};
+
+struct memleak_preinit_pointer {
+ enum memleak_action type;
+ const void *pointer;
+ unsigned long offset;
+ size_t size;
+ unsigned long type_id;
+ int ref_count;
+};
+
+/* Pointer colors, encoded with count and ref_count:
+ * - white - orphan block, i.e. not enough references to it (ref_count >= 1)
+ * - gray - referred at least once and therefore non-orphan (ref_count == 0)
+ * - black - ignore; it doesn't contain references (text section) (ref_count == -1)
+ */
+static inline int color_white(const struct memleak_pointer *pointer)
+{
+ return pointer->count != -1 && pointer->count < pointer->ref_count;
+}
+
+static inline int color_gray(const struct memleak_pointer *pointer)
+{
+ return pointer->ref_count != -1 && pointer->count >= pointer->ref_count;
+}
+
+static inline int color_black(const struct memleak_pointer *pointer)
+{
+ return pointer->ref_count == -1;
+}
+
+/* Tree storing the pointer aliases indexed by size */
+static RADIX_TREE(alias_tree, GFP_ATOMIC);
+/* Tree storing all the possible pointers, indexed by the pointer value */
+static RADIX_TREE(pointer_tree, GFP_ATOMIC);
+/* The list of all allocated pointers */
+static LIST_HEAD(pointer_list);
+/* The list of the gray pointers */
+static LIST_HEAD(gray_list);
+
+static struct kmem_cache *pointer_cache;
+/* Used to avoid recursive call via the kmalloc/kfree functions */
+static spinlock_t memleak_lock = SPIN_LOCK_UNLOCKED;
+static cpumask_t memleak_cpu_mask = CPU_MASK_NONE;
+static DEFINE_MUTEX(memleak_mutex);
+static atomic_t memleak_initialized = ATOMIC_INIT(0);
+static int __initdata preinit_pos;
+static struct memleak_preinit_pointer __initdata preinit_pointers[PREINIT_POINTERS];
+/* last allocated pointer (optimization); protected by memleak_lock */
+static struct memleak_pointer *last_pointer;
+#ifdef CONFIG_DEBUG_FS
+static int reported_leaks;
+#endif
+
+/* pointer flags */
+#define POINTER_ALLOCATED 1
+#define POINTER_ALIASES 2
+
+#ifdef CONFIG_KALLSYMS
+static inline void dump_symbol_name(unsigned long addr)
+{
+ char namebuf[KSYM_NAME_LEN + 1] = "";
+ char *modname;
+ unsigned long symsize;
+ unsigned long offset = 0;
+
+ kallsyms_lookup(addr, &symsize, &offset, &modname, namebuf);
+ printk(KERN_NOTICE " %lx: <%s>\n", addr, namebuf);
+}
+#else
+static inline void dump_symbol_name(unsigned long addr)
+{
+ printk(KERN_NOTICE " %lx\n", addr);
+}
+#endif
+
+#ifdef DEBUG
+static inline void dump_pointer_internals(struct memleak_pointer *pointer)
+{
+ struct memleak_alias *alias;
+ struct hlist_node *elem;
+
+ printk(KERN_NOTICE " size = %d\n", pointer->size);
+ printk(KERN_NOTICE " ref_count = %d\n", pointer->ref_count);
+ printk(KERN_NOTICE " count = %d\n", pointer->count);
+ printk(KERN_NOTICE " aliases:\n");
+ if (pointer->alias_list)
+ hlist_for_each_entry(alias, elem, pointer->alias_list, node)
+ printk(KERN_NOTICE " 0x%lx\n", alias->offset);
+}
+#else
+static inline void dump_pointer_internals(struct memleak_pointer *pointer)
+{ }
+#endif
+
+static void dump_pointer_info(struct memleak_pointer *pointer)
+{
+ int i;
+
+ printk(KERN_NOTICE "kmemleak: pointer 0x%08lx:\n", pointer->pointer);
+ dump_pointer_internals(pointer);
+ printk(KERN_NOTICE " trace:\n");
+ for (i = 0; i < MAX_TRACE; i++) {
+ unsigned long trace = pointer->trace[i];
+
+ if (!trace)
+ break;
+ dump_symbol_name(trace);
+ }
+}
+
+/* Insert an element into the aliases radix tree.
+ * Return 0 on success.
+ */
+static int insert_alias(unsigned long type_id, unsigned long offset)
+{
+ int ret = 0;
+ struct hlist_head *alias_list;
+ struct memleak_alias *alias;
+
+ if (type_id == 0 || offset == 0 || offset >= ml_sizeof(type_id)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ offset &= ~(BYTES_PER_WORD - 1);
+
+ alias_list = radix_tree_lookup(&alias_tree, type_id);
+ if (alias_list) {
+ struct hlist_node *elem;
+
+ hlist_for_each_entry(alias, elem, alias_list, node) {
+ if (alias->offset == offset) {
+ ret = -EEXIST;
+ goto out;
+ }
+ }
+ } else {
+ alias_list = kmalloc(sizeof(*alias_list), GFP_ATOMIC);
+ if (!alias_list)
+ panic("kmemleak: cannot allocate initial memory\n");
+ INIT_HLIST_HEAD(alias_list);
+
+ ret = radix_tree_insert(&alias_tree, type_id, alias_list);
+ if (ret)
+ panic("kmemleak: cannot insert into the aliases radix tree: %d\n", ret);
+ }
+
+ alias = kmalloc(sizeof(*alias), GFP_ATOMIC);
+ if (!alias)
+ panic("kmemleak: cannot allocate initial memory\n");
+ INIT_HLIST_NODE(&alias->node);
+ alias->offset = offset;
+
+ hlist_add_head(&alias->node, alias_list);
+
+ out:
+ return ret;
+}
+
+/* Insert pointer aliases the from the given array */
+void memleak_insert_aliases(struct memleak_offset *ml_off_start,
+ struct memleak_offset *ml_off_end)
+{
+ struct memleak_offset *ml_off;
+ int i = 0;
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ pr_debug("%s(0x%p, 0x%p)\n", __FUNCTION__, ml_off_start, ml_off_end);
+
+ spin_lock_irqsave(&memleak_lock, flags);
+
+ /* do not track the kmemleak allocated pointers */
+ cpu_id = smp_processor_id();
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ BUG();
+
+ /* primary aliases - container_of(member) */
+ for (ml_off = ml_off_start; ml_off < ml_off_end; ml_off++)
+ if (!insert_alias(ml_off->type_id, ml_off->offset))
+ i++;
+ pr_debug("kmemleak: found %d primary alias(es)\n", i);
+
+#ifdef CONFIG_DEBUG_MEMLEAK_SECONDARY_ALIASES
+ /* secondary aliases - container_of(container_of(member)) */
+ for (ml_off = ml_off_start; ml_off < ml_off_end; ml_off++) {
+ struct hlist_head *alias_list;
+ struct memleak_alias *alias;
+ struct hlist_node *elem;
+
+ alias_list = radix_tree_lookup(&alias_tree, ml_off->member_type_id);
+ if (!alias_list)
+ continue;
+
+ hlist_for_each_entry(alias, elem, alias_list, node)
+ if (!insert_alias(ml_off->type_id, ml_off->offset + alias->offset))
+ i++;
+ }
+ pr_debug("kmemleak: found %d alias(es)\n", i);
+#endif
+
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ spin_unlock_irqrestore(&memleak_lock, flags);
+}
+EXPORT_SYMBOL_GPL(memleak_insert_aliases);
+
+static inline struct memleak_pointer *get_cached_pointer(unsigned long ptr)
+{
+ if (!last_pointer || ptr != last_pointer->pointer)
+ last_pointer = radix_tree_lookup(&pointer_tree, ptr);
+ return last_pointer;
+}
+
+static void create_pointer_aliases(struct memleak_pointer *pointer)
+{
+ struct memleak_alias *alias;
+ struct hlist_node *elem;
+ unsigned long ptr = pointer->pointer;
+ int err;
+
+ BUG_ON(pointer->flags & POINTER_ALIASES);
+
+ if (pointer->offset) {
+ err = radix_tree_insert(&pointer_tree, ptr + pointer->offset, pointer);
+ if (err) {
+ dump_stack();
+ panic("kmemleak: cannot insert offset into the pointer radix tree: %d\n", err);
+ }
+ }
+
+ pointer->alias_list = radix_tree_lookup(&alias_tree, pointer->type_id);
+ if (pointer->alias_list) {
+ hlist_for_each_entry(alias, elem, pointer->alias_list, node) {
+ err = radix_tree_insert(&pointer_tree, ptr
+ + pointer->offset + alias->offset,
+ pointer);
+ if (err) {
+ dump_stack();
+ panic("kmemleak: cannot insert alias into the pointer radix tree: %d\n", err);
+ }
+ }
+ }
+
+ pointer->flags |= POINTER_ALIASES;
+}
+
+static void delete_pointer_aliases(struct memleak_pointer *pointer)
+{
+ struct memleak_alias *alias;
+ struct hlist_node *elem;
+ unsigned long ptr = pointer->pointer;
+
+ BUG_ON(!(pointer->flags & POINTER_ALIASES));
+
+ if (pointer->offset)
+ radix_tree_delete(&pointer_tree, ptr + pointer->offset);
+
+ if (pointer->alias_list) {
+ hlist_for_each_entry(alias, elem, pointer->alias_list, node)
+ radix_tree_delete(&pointer_tree,
+ ptr + pointer->offset + alias->offset);
+ pointer->alias_list = NULL;
+ }
+
+ pointer->flags &= ~POINTER_ALIASES;
+}
+
+/* no need for atomic operations since memleak_lock is held anyway */
+static inline void get_pointer(struct memleak_pointer *pointer)
+{
+ pointer->use_count++;
+}
+
+/* called with memleak_lock held for pointer_list modification and
+ * memleak_cpu_mask set to avoid entering memleak_free (and deadlock)
+ */
+static void __put_pointer(struct memleak_pointer *pointer)
+{
+ struct hlist_node *elem, *tmp;
+ struct memleak_scan_area *area;
+
+ if (--pointer->use_count > 0)
+ return;
+
+ /* free the scanning areas */
+ hlist_for_each_entry_safe(area, elem, tmp, &pointer->area_list, node) {
+ hlist_del(elem);
+ kfree(area);
+ }
+
+ list_del(&pointer->pointer_list);
+ kmem_cache_free(pointer_cache, pointer);
+}
+
+static void put_pointer(struct memleak_pointer *pointer)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ spin_lock_irqsave(&memleak_lock, flags);
+ cpu_id = smp_processor_id();
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ BUG();
+
+ __put_pointer(pointer);
+
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ spin_unlock_irqrestore(&memleak_lock, flags);
+}
+
+/* Insert a pointer and its aliases into the pointer radix tree */
+static inline void create_pointer(unsigned long ptr, size_t size, int ref_count)
+{
+ struct memleak_pointer *pointer;
+ int err;
+#ifdef CONFIG_FRAME_POINTER
+ int i;
+ void *frame;
+#endif
+
+ pointer = kmem_cache_alloc(pointer_cache, SLAB_ATOMIC);
+ if (!pointer)
+ panic("kmemleak: cannot allocate a memleak_pointer structure\n");
+
+ last_pointer = pointer;
+
+ INIT_LIST_HEAD(&pointer->pointer_list);
+ INIT_LIST_HEAD(&pointer->gray_list);
+ INIT_HLIST_HEAD(&pointer->area_list);
+ pointer->flags = POINTER_ALLOCATED;
+ pointer->use_count = 0;
+ pointer->pointer = ptr;
+ pointer->offset = 0;
+ pointer->size = size;
+ pointer->type_id = ml_guess_typeid(size); /* type id approximation */
+ pointer->ref_count = ref_count;
+ pointer->count = -1;
+ pointer->alias_list = NULL;
+
+#ifdef CONFIG_FRAME_POINTER
+ frame = __builtin_frame_address(0);
+ for (i = 0; i < MAX_TRACE; i++) {
+ void *stack = task_stack_page(current);
+
+ if (frame < stack || frame > stack + THREAD_SIZE - BYTES_PER_WORD) {
+ pointer->trace[i] = 0;
+ continue;
+ }
+
+ pointer->trace[i] = arch_call_address(frame);
+ frame = arch_prev_frame(frame);
+ /* we don't need the return to do_exit() */
+ if (kstack_end(frame))
+ pointer->trace[i] = 0;
+ }
+#else
+ pointer->trace[0] = (unsigned long)__builtin_return_address(0);
+#endif
+
+ err = radix_tree_insert(&pointer_tree, ptr, pointer);
+ if (err) {
+ dump_stack();
+ if (err == -EEXIST) {
+ printk(KERN_NOTICE "Existing pointer:\n");
+ pointer = radix_tree_lookup(&pointer_tree, ptr);
+ dump_pointer_info(pointer);
+ }
+ panic("kmemleak: cannot insert 0x%lx into the pointer radix tree: %d\n",
+ ptr, err);
+ }
+
+ list_add_tail(&pointer->pointer_list, &pointer_list);
+ get_pointer(pointer);
+}
+
+/* Remove a pointer and its aliases from the pointer radix tree */
+static inline void delete_pointer(unsigned long ptr)
+{
+ struct memleak_pointer *pointer;
+
+ pointer = radix_tree_delete(&pointer_tree, ptr);
+ if (!pointer) {
+ dump_stack();
+ printk(KERN_WARNING "kmemleak: freeing unknown pointer value 0x%08lx\n", ptr);
+ return;
+ }
+ if (pointer->pointer != ptr) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: freeing pointer by alias 0x%08lx\n", ptr);
+ }
+
+ BUG_ON(!(pointer->flags & POINTER_ALLOCATED));
+
+ if (last_pointer && ptr == last_pointer->pointer)
+ last_pointer = NULL;
+
+#ifdef CONFIG_DEBUG_MEMLEAK_ORPHAN_FREEING
+ if (color_white(pointer)) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ printk(KERN_WARNING "kmemleak: freeing orphan pointer 0x%08lx\n", ptr);
+ }
+#endif
+
+ if (pointer->flags & POINTER_ALIASES)
+ delete_pointer_aliases(pointer);
+ pointer->flags &= ~POINTER_ALLOCATED;
+ pointer->pointer = 0;
+ __put_pointer(pointer);
+}
+
+/* Re-create the pointer aliases according to the new size/offset
+ * information */
+static inline void unpad_pointer(unsigned long ptr, unsigned long offset,
+ size_t size)
+{
+ struct memleak_pointer *pointer;
+
+ pointer = get_cached_pointer(ptr);
+ if (!pointer) {
+ dump_stack();
+ panic("kmemleak: resizing unknown pointer value 0x%08lx\n", ptr);
+ }
+ if (pointer->pointer != ptr) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: resizing pointer by alias 0x%08lx\n", ptr);
+ }
+ if (offset + size > pointer->size) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: new boundaries exceed block 0x%08lx\n", ptr);
+ }
+
+ if (offset == pointer->offset && size == pointer->size)
+ return;
+
+ if (pointer->flags & POINTER_ALIASES)
+ delete_pointer_aliases(pointer);
+
+ pointer->offset = offset;
+ pointer->size = size;
+}
+
+/* Make a pointer permanently gray (false positive) */
+static inline void make_gray_pointer(unsigned long ptr)
+{
+ struct memleak_pointer *pointer;
+
+ pointer = get_cached_pointer(ptr);
+ if (!pointer) {
+ dump_stack();
+ panic("kmemleak: graying unknown pointer value 0x%08lx\n", ptr);
+ }
+ if (pointer->pointer != ptr) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: graying pointer by alias 0x%08lx\n", ptr);
+ }
+
+ pointer->ref_count = 0;
+}
+
+/* Mark the pointer as black */
+static inline void make_black_pointer(unsigned long ptr)
+{
+ struct memleak_pointer *pointer;
+
+ pointer = get_cached_pointer(ptr);
+ if (!pointer) {
+ dump_stack();
+ panic("kmemleak: blacking unknown pointer value 0x%08lx\n", ptr);
+ }
+ if (pointer->pointer != ptr) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: blacking pointer by alias 0x%08lx\n", ptr);
+ }
+
+ pointer->ref_count = -1;
+}
+
+/* Add a scanning area to the pointer */
+static inline void add_scan_area(unsigned long ptr, unsigned long offset, size_t length)
+{
+ struct memleak_pointer *pointer;
+ struct memleak_scan_area *area;
+
+ pointer = get_cached_pointer(ptr);
+ if (!pointer) {
+ dump_stack();
+ panic("kmemleak: adding scan area to unknown pointer value 0x%08lx\n", ptr);
+ }
+ if (pointer->pointer != ptr) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: adding scan area to pointer by alias 0x%08lx\n", ptr);
+ }
+ if (offset + length > pointer->size) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: scan area larger than block 0x%08lx\n", ptr);
+ }
+
+ area = kmalloc(sizeof(*area), GFP_ATOMIC);
+ if (!area)
+ panic("kmemleak: cannot allocate a scan area\n");
+
+ INIT_HLIST_NODE(&area->node);
+ area->offset = offset;
+ area->length = length;
+
+ hlist_add_head(&area->node, &pointer->area_list);
+}
+
+/* Re-create the pointer aliases according to the new type id */
+static inline void change_type_id(unsigned long ptr, unsigned long type_id)
+{
+ struct memleak_pointer *pointer;
+
+ pointer = get_cached_pointer(ptr);
+ if (!pointer) {
+ dump_stack();
+ panic("kmemleak: changing type of unknown pointer value 0x%08lx\n", ptr);
+ }
+ if (pointer->pointer != ptr) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: changing type of pointer by alias 0x%08lx\n", ptr);
+ }
+ if (ml_sizeof(type_id) > pointer->size) {
+ dump_stack();
+ dump_pointer_info(pointer);
+ panic("kmemleak: new type larger than block 0x%08lx\n", ptr);
+ }
+
+ if (!type_id || type_id == pointer->type_id)
+ return;
+
+ if (pointer->flags & POINTER_ALIASES)
+ delete_pointer_aliases(pointer);
+
+ pointer->type_id = type_id;
+}
+
+/* Allocation function hook */
+void memleak_alloc(const void *ptr, size_t size, int ref_count)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. After disabling the interrupts, the
+ * only calls to this function on the same CPU should be from
+ * kmemleak itself and we ignore them. Calls from other CPU's
+ * would wait on the spin_lock.
+ */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p, %d, %d)\n", __FUNCTION__, ptr, size, ref_count);
+
+ if (!atomic_read(&memleak_initialized)) {
+ /* no need for SMP locking since this block is
+ * executed before the other CPUs are started */
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_ALLOC;
+ pointer->pointer = ptr;
+ pointer->size = size;
+ pointer->ref_count = ref_count;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ create_pointer((unsigned long)ptr, size, ref_count);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(memleak_alloc);
+
+/* Freeing function hook */
+void memleak_free(const void *ptr)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. See memleak_alloc() for an explanation */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p)\n", __FUNCTION__, ptr);
+
+ if (!atomic_read(&memleak_initialized)) {
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_FREE;
+ pointer->pointer = ptr;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ delete_pointer((unsigned long)ptr);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(memleak_free);
+
+/* Change the size and location information of an allocated memory
+ * block (this is needed for allocations padding the object) */
+void memleak_padding(const void *ptr, unsigned long offset, size_t size)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. See memleak_alloc() for an explanation */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p, %d)\n", __FUNCTION__, ptr, size);
+
+ if (!atomic_read(&memleak_initialized)) {
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_PADDING;
+ pointer->pointer = ptr;
+ pointer->offset = offset;
+ pointer->size = size;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ unpad_pointer((unsigned long)ptr, offset, size);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL(memleak_padding);
+
+/* Mark a pointer as a false positive */
+void memleak_not_leak(const void *ptr)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. See memleak_alloc() for an explanation */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p)\n", __FUNCTION__, ptr);
+
+ if (!atomic_read(&memleak_initialized)) {
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_NOT_LEAK;
+ pointer->pointer = ptr;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ make_gray_pointer((unsigned long)ptr);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL(memleak_not_leak);
+
+/* Ignore this memory block */
+void memleak_ignore(const void *ptr)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. See memleak_alloc() for an explanation */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p)\n", __FUNCTION__, ptr);
+
+ if (!atomic_read(&memleak_initialized)) {
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_IGNORE;
+ pointer->pointer = ptr;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ make_black_pointer((unsigned long)ptr);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL(memleak_ignore);
+
+/* Add a scanning area to a pointer */
+void memleak_scan_area(const void *ptr, unsigned long offset, size_t length)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. See memleak_alloc() for an explanation */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p)\n", __FUNCTION__, ptr);
+
+ if (!atomic_read(&memleak_initialized)) {
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_SCAN_AREA;
+ pointer->pointer = ptr;
+ pointer->offset = offset;
+ pointer->size = length;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ add_scan_area((unsigned long)ptr, offset, length);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL(memleak_scan_area);
+
+/* Change the type id of an allocated memory block */
+void memleak_typeid_raw(const void *ptr, unsigned long type_id)
+{
+ unsigned long flags;
+ unsigned int cpu_id;
+
+ if (!ptr)
+ return;
+
+ local_irq_save(flags);
+ cpu_id = get_cpu();
+
+ /* avoid recursive calls. See memleak_alloc() for an explanation */
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ goto out;
+
+ pr_debug("%s(0x%p, %ld)\n", __FUNCTION__, ptr, type_id);
+
+ if (!atomic_read(&memleak_initialized)) {
+ struct memleak_preinit_pointer *pointer;
+
+ BUG_ON(cpu_id != 0);
+
+ if (preinit_pos >= PREINIT_POINTERS)
+ panic("kmemleak: preinit pointers buffer overflow\n");
+ pointer = &preinit_pointers[preinit_pos++];
+
+ pointer->type = MEMLEAK_TYPEID;
+ pointer->pointer = ptr;
+ pointer->type_id = type_id;
+
+ goto unmask;
+ }
+
+ spin_lock(&memleak_lock);
+ change_type_id((unsigned long)ptr, type_id);
+ spin_unlock(&memleak_lock);
+
+ unmask:
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ out:
+ put_cpu_no_resched();
+ local_irq_restore(flags);
+}
+EXPORT_SYMBOL(memleak_typeid_raw);
+
+/* Scan a block of memory (exclusive range) for pointers and move
+ * those found to the gray list. This function is called with
+ * memleak_lock held
+ */
+static void __scan_block(void *_start, void *_end)
+{
+ unsigned long *ptr;
+ unsigned long *start = (unsigned long *)ALIGN((unsigned long)_start,
+ BYTES_PER_WORD);
+ unsigned long *end = _end;
+
+ for (ptr = start; ptr < end; ptr++) {
+ struct memleak_pointer *pointer =
+ radix_tree_lookup(&pointer_tree,
+ (*ptr) & ~(BYTES_PER_WORD - 1));
+ if (!pointer)
+ continue;
+ if (!color_white(pointer))
+ continue;
+
+ pointer->count++;
+ /* this can happen during the grey_list traversal */
+ if (color_gray(pointer)) {
+ get_pointer(pointer);
+ list_add_tail_rcu(&pointer->gray_list, &gray_list);
+ }
+ }
+}
+
+static void scan_block(void *start, void *end)
+{
+ unsigned long flags;
+ void *s, *e;
+
+ s = start;
+ while (s < end) {
+ e = s + SCAN_BLOCK_SIZE;
+
+ spin_lock_irqsave(&memleak_lock, flags);
+ __scan_block(s, e < end ? e : end);
+ spin_unlock_irqrestore(&memleak_lock, flags);
+
+ s = e;
+ }
+}
+
+/* Scan a memory block represented by a memleak_pointer */
+static inline void scan_pointer(struct memleak_pointer *pointer)
+{
+ struct memleak_scan_area *area;
+ struct hlist_node *elem;
+ unsigned long flags;
+
+ spin_lock_irqsave(&memleak_lock, flags);
+
+ /* freed pointer */
+ if (!pointer->pointer)
+ goto out;
+
+ if (hlist_empty(&pointer->area_list))
+ __scan_block((void *)pointer->pointer,
+ (void *)(pointer->pointer + pointer->size));
+ else
+ hlist_for_each_entry(area, elem, &pointer->area_list, node) {
+ unsigned long ptr = pointer->pointer + area->offset;
+
+ __scan_block((void *)ptr, (void *)(ptr + area->length));
+ }
+
+ out:
+ spin_unlock_irqrestore(&memleak_lock, flags);
+}
+
+/* Scan the memory and print the orphan pointers */
+static void memleak_scan(void)
+{
+ unsigned long flags;
+ struct memleak_pointer *pointer, *tmp;
+#ifdef CONFIG_DEBUG_MEMLEAK_TASK_STACKS
+ struct task_struct *task;
+#endif
+ int i;
+ unsigned int cpu_id;
+
+ /* initialize pointers (make them white) and build the initial
+ * gray list. Note that create_pointer_aliases might allocate
+ * memory */
+ spin_lock_irqsave(&memleak_lock, flags);
+ cpu_id = smp_processor_id();
+ if (cpu_test_and_set(cpu_id, memleak_cpu_mask))
+ BUG();
+
+ list_for_each_entry(pointer, &pointer_list, pointer_list) {
+ /* lazy insertion of the pointer aliases into the radix tree */
+ if ((pointer->flags & POINTER_ALLOCATED)
+ && !(pointer->flags & POINTER_ALIASES))
+ create_pointer_aliases(pointer);
+
+ pointer->count = 0;
+ if (color_gray(pointer)) {
+ get_pointer(pointer);
+ list_add_tail(&pointer->gray_list, &gray_list);
+ }
+ }
+
+ cpu_clear(cpu_id, memleak_cpu_mask);
+ spin_unlock_irqrestore(&memleak_lock, flags);
+
+ /* data/bss scanning */
+ scan_block(_sdata, _edata);
+ scan_block(__bss_start, __bss_stop);
+
+#ifdef CONFIG_SMP
+ /* per-cpu scanning */
+ for (i = 0; i < NR_CPUS; i++)
+ scan_block(__per_cpu_offset[i] + __per_cpu_start,
+ __per_cpu_offset[i] + __per_cpu_end);
+#endif
+
+ /* mem_map scanning */
+ for_each_online_node(i) {
+ struct page *page, *end;
+
+ page = NODE_MEM_MAP(i);
+ end = page + NODE_DATA(i)->node_spanned_pages;
+
+ scan_block(page, end);
+ }
+
+#ifdef CONFIG_DEBUG_MEMLEAK_TASK_STACKS
+ read_lock(&tasklist_lock);
+ for_each_process(task)
+ scan_block(task_stack_page(task),
+ task_stack_page(task) + THREAD_SIZE);
+ read_unlock(&tasklist_lock);
+#endif
+
+ /* gray_list scanning. RCU is needed because new elements can
+ * be added to the list during scanning */
+ rcu_read_lock();
+ list_for_each_entry_rcu(pointer, &gray_list, gray_list)
+ scan_pointer(pointer);
+ rcu_read_unlock();
+
+ /* empty the gray list. It needs the "safe" version because
+ * put_pointer() can free the structure */
+ list_for_each_entry_safe(pointer, tmp, &gray_list, gray_list) {
+ list_del(&pointer->gray_list);
+ put_pointer(pointer);
+ }
+}
+
+#ifdef CONFIG_DEBUG_FS
+static void *memleak_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ struct memleak_pointer *pointer;
+ loff_t n = *pos;
+ unsigned long flags;
+
+ mutex_lock(&memleak_mutex);
+
+ if (!n) {
+ memleak_scan();
+ reported_leaks = 0;
+ }
+ if (reported_leaks >= CONFIG_DEBUG_MEMLEAK_REPORTS_NR)
+ return NULL;
+
+ spin_lock_irqsave(&memleak_lock, flags);
+
+ list_for_each_entry(pointer, &pointer_list, pointer_list)
+ if (!n--) {
+ get_pointer(pointer);
+ goto out;
+ }
+ pointer = NULL;
+
+ out:
+ spin_unlock_irqrestore(&memleak_lock, flags);
+ return pointer;
+}
+
+static void *memleak_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+ struct list_head *n;
+ struct memleak_pointer *next = NULL;
+ unsigned long flags;
+
+ ++(*pos);
+ if (reported_leaks >= CONFIG_DEBUG_MEMLEAK_REPORTS_NR)
+ return NULL;
+
+ spin_lock_irqsave(&memleak_lock, flags);
+
+ n = ((struct memleak_pointer *)v)->pointer_list.next;
+ if (n != &pointer_list) {
+ next = list_entry(n, struct memleak_pointer, pointer_list);
+ get_pointer(next);
+ }
+
+ spin_unlock_irqrestore(&memleak_lock, flags);
+
+ put_pointer(v);
+ return next;
+}
+
+static void memleak_seq_stop(struct seq_file *seq, void *v)
+{
+ if (v)
+ put_pointer(v);
+ mutex_unlock(&memleak_mutex);
+}
+
+static int memleak_seq_show(struct seq_file *seq, void *v)
+{
+ const struct memleak_pointer *pointer = v;
+#ifdef CONFIG_KALLSYMS
+ char namebuf[KSYM_NAME_LEN + 1] = "";
+ char *modname;
+ unsigned long symsize;
+ unsigned long offset = 0;
+#endif
+ int i;
+
+ if (!color_white(pointer))
+ return 0;
+ /* freed in the meantime (false positive) or just allocated */
+ if (!(pointer->flags & POINTER_ALLOCATED)
+ || !(pointer->flags & POINTER_ALIASES))
+ return 0;
+
+ reported_leaks++;
+ seq_printf(seq, "orphan pointer 0x%08lx (size %d):\n",
+ pointer->pointer, pointer->size);
+
+ for (i = 0; i < MAX_TRACE; i++) {
+ unsigned long trace = pointer->trace[i];
+ if (!trace)
+ break;
+
+#ifdef CONFIG_KALLSYMS
+ kallsyms_lookup(trace, &symsize, &offset, &modname, namebuf);
+ seq_printf(seq, " %lx: <%s>\n", trace, namebuf);
+#else
+ seq_printf(seq, " %lx\n", trace);
+#endif
+ }
+
+ return 0;
+}
+
+static struct seq_operations memleak_seq_ops = {
+ .start = memleak_seq_start,
+ .next = memleak_seq_next,
+ .stop = memleak_seq_stop,
+ .show = memleak_seq_show,
+};
+
+static int memleak_seq_open(struct inode *inode, struct file *file)
+{
+ return seq_open(file, &memleak_seq_ops);
+}
+
+static struct file_operations memleak_fops = {
+ .owner = THIS_MODULE,
+ .open = memleak_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = seq_release,
+};
+#endif /* CONFIG_DEBUG_FS */
+
+/* KMemLeak initialization. Set up the radix tree for the pointer aliases */
+void __init memleak_init(void)
+{
+ int i;
+ unsigned long flags;
+
+ pointer_cache = kmem_cache_create("pointer_cache", sizeof(struct memleak_pointer),
+ 0, SLAB_PANIC, NULL, NULL);
+ if (!pointer_cache)
+ panic("kmemleak: cannot create the pointer cache\n");
+
+ memleak_insert_aliases(__memleak_offsets_start, __memleak_offsets_end);
+
+ /* no need to hold the spinlock as SMP is not initialized
+ * yet. Holding it here would lead to a deadlock */
+ local_irq_save(flags);
+
+ atomic_set(&memleak_initialized, 1);
+
+ /* execute the buffered memleak actions */
+ pr_debug("kmemleak: %d preinit actions\n", preinit_pos);
+ for (i = 0; i < preinit_pos; i++) {
+ struct memleak_preinit_pointer *pointer = &preinit_pointers[i];
+
+ switch (pointer->type) {
+ case MEMLEAK_ALLOC:
+ memleak_alloc(pointer->pointer, pointer->size,
+ pointer->ref_count);
+ break;
+ case MEMLEAK_FREE:
+ memleak_free(pointer->pointer);
+ break;
+ case MEMLEAK_PADDING:
+ memleak_padding(pointer->pointer, pointer->offset,
+ pointer->size);
+ break;
+ case MEMLEAK_NOT_LEAK:
+ memleak_not_leak(pointer->pointer);
+ break;
+ case MEMLEAK_IGNORE:
+ memleak_ignore(pointer->pointer);
+ break;
+ case MEMLEAK_SCAN_AREA:
+ memleak_scan_area(pointer->pointer,
+ pointer->offset, pointer->size);
+ break;
+ case MEMLEAK_TYPEID:
+ memleak_typeid_raw(pointer->pointer, pointer->type_id);
+ break;
+ default:
+ BUG();
+ }
+ }
+
+ local_irq_restore(flags);
+
+ printk(KERN_INFO "Kernel memory leak detector initialized\n");
+}
+
+/* Late initialization function */
+int __init memleak_late_init(void)
+{
+#ifdef CONFIG_DEBUG_FS
+ struct dentry *dentry;
+
+ dentry = debugfs_create_file("memleak", S_IRUGO, NULL, NULL,
+ &memleak_fops);
+ if (!dentry)
+ return -ENOMEM;
+#endif
+ pr_debug("kmemleak: late initialization completed\n");
+
+ return 0;
+}
+late_initcall(memleak_late_init);

2006-08-12 22:00:08

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 02/10] Some documentation for kmemleak

From: Catalin Marinas <[email protected]>

Signed-off-by: Catalin Marinas <[email protected]>
---

Documentation/kmemleak.txt | 157 ++++++++++++++++++++++++++++++++++++++++++++
1 files changed, 157 insertions(+), 0 deletions(-)

diff --git a/Documentation/kmemleak.txt b/Documentation/kmemleak.txt
new file mode 100644
index 0000000..4d2f608
--- /dev/null
+++ b/Documentation/kmemleak.txt
@@ -0,0 +1,157 @@
+Kernel Memory Leak Detector
+===========================
+
+
+Introduction
+------------
+
+Kmemleak provides a way of detecting possible kernel memory leaks in a
+way similar to a tracing garbage collector
+(http://en.wikipedia.org/wiki/Garbage_collection_%28computer_science%29#Tracing_garbage_collectors),
+with the difference that the orphan pointers are not freed but only
+reported via /sys/kernel/debug/memleak. A similar method is used by
+the Valgrind tool (memcheck --leak-check) to detect the memory leaks
+in user-space applications.
+
+
+Usage
+-----
+
+CONFIG_DEBUG_MEMLEAK has to be enabled. For additional config options,
+look in:
+
+ -> Kernel hacking
+ -> Kernel debugging
+ -> Debug slab memory allocations
+ -> Kernel memory leak detector
+
+To display the possible memory leaks:
+
+ # mount -t debugfs nodev /sys/kernel/debug/
+ # cat /sys/kernel/debug/memleak
+
+In order to reduce the run-time overhead, memory scanning is only
+performed when reading the /sys/kernel/debug/memleak file.
+
+
+Basic Algorithm
+---------------
+
+The memory allocations via kmalloc, vmalloc, kmem_cache_alloc and
+friends are tracked and the pointers, together with additional
+information like size and stack trace, are stored in a radix tree. The
+corresponding freeing function calls are tracked and the pointers
+removed from the radix tree.
+
+An allocated block of memory is considered orphan if a pointer to its
+start address or to an alias (pointer aliases are explained later)
+cannot be found by scanning the memory (including saved
+registers). This means that there might be no way for the kernel to
+pass the address of the allocated block to a freeing function and
+therefore the block is considered a leak.
+
+The scanning algorithm steps:
+
+ 1. mark all pointers as white (remaining white pointers will later
+ be considered orphan)
+ 2. scan the memory starting with the data section and stacks,
+ checking the values against the addresses stored in the radix
+ tree. If a white pointer is found, it is added to the grey list
+ 3. scan the grey pointers for matching addresses (some white
+ pointers can become grey and added at the end of the grey list)
+ until the grey set is finished
+ 4. the remaining white pointers are considered orphan and reported
+ via /sys/kernel/debug/memleak
+
+
+Improvements
+------------
+
+Because the Linux kernel calculates many pointers at run-time via the
+container_of macro (see the lists implementation), a lot of false
+positives would be reported. This tool re-writes the container_of
+macro so that the offset and type information is stored in the
+.init.memleak_offsets section. The memleak_init() function creates a
+radix tree with corresponding offsets for every encountered block
+type. The memory allocations hook stores the pointer address together
+with its aliases based on the type of the allocated block.
+
+While one level of offsets should be enough for most cases, a second
+level, i.e. container_of(container_of(...)), can be enabled via the
+configuration options (one false positive is the "struct socket_alloc"
+allocation in the sock_alloc_inode() function).
+
+Some allocated memory blocks have pointers stored in the kernel's
+internal data structures and they cannot be detected as orphans. To
+avoid this, kmemleak can also store the number of values equal to the
+pointer (or aliases) that need to be found so that the block is not
+considered a leak. One example is __vmalloc().
+
+
+Limitations and Drawbacks
+-------------------------
+
+The biggest drawback is the reduced performance of memory allocation
+and freeing. To avoid other penalties, the memory scanning is only
+performed when the /sys/kernel/debug/memleak file is read. Anyway,
+this tool is intended for debugging purposes where the performance
+might not be the most important requirement.
+
+Kmemleak currently approximates the type id using the sizeof()
+compiler built-in function. This is not accurate and can lead to false
+negatives. The aim is to gradually change the kernel and kmemleak to
+do more precise type identification.
+
+Another source of false negatives is the data stored in non-pointer
+values. Together with the more precise type identification, kmemleak
+could only scan the pointer members in the allocated structures.
+
+The tool can report false positives. These are cases where an
+allocated block doesn't need to be freed (some cases in the init_call
+functions), the pointer is calculated by other methods than the
+container_of macro or the pointer is stored in a location not scanned
+by kmemleak. If the "member" argument in the offsetof(type, member)
+call is not constant, kmemleak considers the offset as zero since it
+cannot be determined at compilation time.
+
+Page allocations and ioremap are not tracked. Only the ARM and i386
+architectures are currently supported.
+
+
+Kmemleak API
+------------
+
+See the include/linux/memleak.h header for the functions prototype.
+
+memleak_init - initialize kmemleak
+memleak_alloc - notify of a memory block allocation
+memleak_free - notify of a memory block freeing
+memleak_padding - mark the boundaries of the data inside the block
+memleak_not_leak - mark a pointer as not a leak
+memleak_ignore - do not scan or report a pointer as leak
+memleak_scan_area - add scan areas inside a memory block
+memleak_insert_aliases - add aliases for a given type
+memleak_erase - erase an old value in a pointer variable
+memleak_typeid_raw - set the typeid for an allocated block
+memleak_container - statically declare a pointer alias
+memleak_typeid - set the typeid for an allocated block (takes
+ a type rather than typeid as argument)
+
+
+Dealing with false positives/negatives
+--------------------------------------
+
+To reduce the false negatives, kmemleak provides the memleak_ignore,
+memleak_scan_area and memleak_erase functions. The task stacks also
+increase the amount of false negatives and their scanning is not
+enabled by default.
+
+To eliminate the false positives caused by code allocating a different
+size from the object one (either for alignment or for extra memory
+after the end of the structure), kmemleak provides the memleak_padding
+and memleak_typeid functions.
+
+For pointers known not to be leaks, kmemleak provides the
+memleak_not_leak function. The memleak_ignore could also be used if
+the memory block is known not to contain other pointers as it will no
+longer be scanned.

2006-08-12 22:00:54

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 03/10] Add the memory allocation/freeing hooks for kmemleak

From: Catalin Marinas <[email protected]>

This patch adds the callbacks to memleak_(alloc|free) functions from
kmalloc/kfree, kmem_cache_(alloc|free), vmalloc/vfree etc.

Signed-off-by: Catalin Marinas <[email protected]>
---

include/linux/slab.h | 6 ++++++
mm/page_alloc.c | 2 ++
mm/slab.c | 21 +++++++++++++++++++--
mm/vmalloc.c | 22 ++++++++++++++++++++--
4 files changed, 47 insertions(+), 4 deletions(-)

diff --git a/include/linux/slab.h b/include/linux/slab.h
index 45ad55b..e3f5945 100644
--- a/include/linux/slab.h
+++ b/include/linux/slab.h
@@ -133,6 +133,8 @@ #endif
*/
static inline void *kmalloc(size_t size, gfp_t flags)
{
+#ifndef CONFIG_DEBUG_MEMLEAK
+ /* this block removes the size information needed by kmemleak */
if (__builtin_constant_p(size)) {
int i = 0;
#define CACHE(x) \
@@ -151,6 +153,7 @@ found:
malloc_sizes[i].cs_dmacachep :
malloc_sizes[i].cs_cachep, flags);
}
+#endif
return __kmalloc(size, flags);
}

@@ -163,6 +166,8 @@ extern void *__kzalloc(size_t, gfp_t);
*/
static inline void *kzalloc(size_t size, gfp_t flags)
{
+#ifndef CONFIG_DEBUG_MEMLEAK
+ /* this block removes the size information needed by kmemleak */
if (__builtin_constant_p(size)) {
int i = 0;
#define CACHE(x) \
@@ -181,6 +186,7 @@ found:
malloc_sizes[i].cs_dmacachep :
malloc_sizes[i].cs_cachep, flags);
}
+#endif
return __kzalloc(size, flags);
}

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 54a4f53..bdf6445 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2450,6 +2450,8 @@ void *__init alloc_large_system_hash(con
if (_hash_mask)
*_hash_mask = (1 << log2qty) - 1;

+ memleak_alloc(table, size, 1);
+
return table;
}

diff --git a/mm/slab.c b/mm/slab.c
index 21ba060..42c90aa 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -2446,6 +2446,9 @@ static struct slab *alloc_slabmgmt(struc
/* Slab management obj is off-slab. */
slabp = kmem_cache_alloc_node(cachep->slabp_cache,
local_flags, nodeid);
+ /* only scan the list member to avoid false negatives */
+ memleak_scan_area(slabp, offsetof(struct slab, list),
+ sizeof(struct list_head));
if (!slabp)
return NULL;
} else {
@@ -2986,6 +2989,8 @@ #endif
STATS_INC_ALLOCMISS(cachep);
objp = cache_alloc_refill(cachep, flags);
}
+ /* avoid false negatives */
+ memleak_erase(&ac->entry[ac->avail]);
return objp;
}

@@ -3002,6 +3007,7 @@ static __always_inline void *__cache_all
local_irq_restore(save_flags);
objp = cache_alloc_debugcheck_after(cachep, flags, objp,
caller);
+ memleak_alloc(objp, obj_size(cachep), 1);
prefetchw(objp);
return objp;
}
@@ -3193,6 +3199,7 @@ static inline void __cache_free(struct k
struct array_cache *ac = cpu_cache_get(cachep);

check_irq_off();
+ memleak_free(objp);
objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0));

if (cache_free_alien(cachep, objp))
@@ -3312,6 +3319,7 @@ void *kmem_cache_alloc_node(struct kmem_

ptr = cache_alloc_debugcheck_after(cachep, flags, ptr,
__builtin_return_address(0));
+ memleak_alloc(ptr, obj_size(cachep), 1);

return ptr;
}
@@ -3320,11 +3328,15 @@ EXPORT_SYMBOL(kmem_cache_alloc_node);
void *kmalloc_node(size_t size, gfp_t flags, int node)
{
struct kmem_cache *cachep;
+ void *ptr;

cachep = kmem_find_general_cachep(size, flags);
if (unlikely(cachep == NULL))
return NULL;
- return kmem_cache_alloc_node(cachep, flags, node);
+ ptr = kmem_cache_alloc_node(cachep, flags, node);
+ memleak_padding(ptr, 0, size);
+ memleak_typeid_raw(ptr, ml_guess_typeid(size));
+ return ptr;
}
EXPORT_SYMBOL(kmalloc_node);
#endif
@@ -3339,6 +3351,7 @@ static __always_inline void *__do_kmallo
void *caller)
{
struct kmem_cache *cachep;
+ void *ptr;

/* If you want to save a few bytes .text space: replace
* __ with kmem_.
@@ -3348,7 +3361,11 @@ static __always_inline void *__do_kmallo
cachep = __find_general_cachep(size, flags);
if (unlikely(cachep == NULL))
return NULL;
- return __cache_alloc(cachep, flags, caller);
+ ptr = __cache_alloc(cachep, flags, caller);
+ memleak_padding(ptr, 0, size);
+ memleak_typeid_raw(ptr, ml_guess_typeid(size));
+
+ return ptr;
}


diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index 266162d..a4d319f 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -364,6 +364,9 @@ void __vunmap(void *addr, int deallocate
void vfree(void *addr)
{
BUG_ON(in_interrupt());
+
+ memleak_free(addr);
+
__vunmap(addr, 1);
}
EXPORT_SYMBOL(vfree);
@@ -463,7 +466,14 @@ fail:

void *__vmalloc_area(struct vm_struct *area, gfp_t gfp_mask, pgprot_t prot)
{
- return __vmalloc_area_node(area, gfp_mask, prot, -1);
+ void *addr = __vmalloc_area_node(area, gfp_mask, prot, -1);
+
+ /* this needs ref_count = 2 since vm_struct also contains a
+ * pointer to this address. The guard page is also subtracted
+ * from the size */
+ memleak_alloc(addr, area->size - PAGE_SIZE, 2);
+
+ return addr;
}

/**
@@ -482,6 +492,8 @@ void *__vmalloc_node(unsigned long size,
int node)
{
struct vm_struct *area;
+ void *addr;
+ unsigned long real_size = size;

size = PAGE_ALIGN(size);
if (!size || (size >> PAGE_SHIFT) > num_physpages)
@@ -491,7 +503,13 @@ void *__vmalloc_node(unsigned long size,
if (!area)
return NULL;

- return __vmalloc_area_node(area, gfp_mask, prot, node);
+ addr = __vmalloc_area_node(area, gfp_mask, prot, node);
+
+ /* this needs ref_count = 2 since the vm_struct also contains
+ a pointer to this address */
+ memleak_alloc(addr, real_size, 2);
+
+ return addr;
}
EXPORT_SYMBOL(__vmalloc_node);

2006-08-12 22:01:13

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 07/10] Remove some of the kmemleak false positives

From: Catalin Marinas <[email protected]>

There are allocations for which the main pointer cannot be found but they
are not memory leaks. This patch fixes some of them.

Signed-off-by: Catalin Marinas <[email protected]>
---

drivers/base/platform.c | 3 +++
drivers/hwmon/w83627hf.c | 4 ++++
drivers/scsi/hosts.c | 3 +++
fs/ext3/dir.c | 3 +++
ipc/util.c | 6 ++++++
kernel/params.c | 8 +++++++-
mm/slab.c | 4 ++++
net/core/dev.c | 6 ++++++
net/core/skbuff.c | 3 +++
net/ipv4/netfilter/ip_conntrack_core.c | 5 +++++
net/sched/sch_generic.c | 6 ++++++
11 files changed, 50 insertions(+), 1 deletions(-)

diff --git a/drivers/base/platform.c b/drivers/base/platform.c
index 2b8755d..1521fe4 100644
--- a/drivers/base/platform.c
+++ b/drivers/base/platform.c
@@ -166,6 +166,9 @@ struct platform_device *platform_device_
struct platform_object *pa;

pa = kzalloc(sizeof(struct platform_object) + strlen(name), GFP_KERNEL);
+ /* kmemleak cannot guess the object type because the block
+ * size is different from the object size */
+ memleak_typeid(pa, struct platform_object);
if (pa) {
strcpy(pa->name, name);
pa->pdev.name = pa->name;
diff --git a/drivers/hwmon/w83627hf.c b/drivers/hwmon/w83627hf.c
index 79368d5..f952f02 100644
--- a/drivers/hwmon/w83627hf.c
+++ b/drivers/hwmon/w83627hf.c
@@ -1065,6 +1065,10 @@ static int w83627hf_detect(struct i2c_ad
err = -ENOMEM;
goto ERROR1;
}
+ /* the pointer to member is stored but the code doesn't use
+ * container_of for access and the alias need to be
+ * explicitely declared here */
+ memleak_container(struct w83627hf_data, client);

new_client = &data->client;
i2c_set_clientdata(new_client, data);
diff --git a/drivers/scsi/hosts.c b/drivers/scsi/hosts.c
index dfcb96f..9516d37 100644
--- a/drivers/scsi/hosts.c
+++ b/drivers/scsi/hosts.c
@@ -297,6 +297,9 @@ struct Scsi_Host *scsi_host_alloc(struct
shost = kzalloc(sizeof(struct Scsi_Host) + privsize, gfp_mask);
if (!shost)
return NULL;
+ /* kmemleak cannot guess the object type because the block
+ * size is different from the object size */
+ memleak_typeid(shost, struct Scsi_Host);

spin_lock_init(&shost->default_lock);
scsi_assign_lock(shost, &shost->default_lock);
diff --git a/fs/ext3/dir.c b/fs/ext3/dir.c
index fbb0d4e..a34814d 100644
--- a/fs/ext3/dir.c
+++ b/fs/ext3/dir.c
@@ -346,6 +346,9 @@ int ext3_htree_store_dirent(struct file
new_fn = kmalloc(len, GFP_KERNEL);
if (!new_fn)
return -ENOMEM;
+ /* kmemleak cannot guess the object type because the block
+ * size is different from the object size */
+ memleak_typeid(new_fn, struct fname);
memset(new_fn, 0, len);
new_fn->hash = hash;
new_fn->minor_hash = minor_hash;
diff --git a/ipc/util.c b/ipc/util.c
index 67b6d17..17cc294 100644
--- a/ipc/util.c
+++ b/ipc/util.c
@@ -388,6 +388,9 @@ void* ipc_rcu_alloc(int size)
*/
if (rcu_use_vmalloc(size)) {
out = vmalloc(HDRLEN_VMALLOC + size);
+ /* the stored pointer is different from the address of
+ * the allocated block because of padding */
+ memleak_padding(out, HDRLEN_VMALLOC, size);
if (out) {
out += HDRLEN_VMALLOC;
container_of(out, struct ipc_rcu_hdr, data)->is_vmalloc = 1;
@@ -395,6 +398,9 @@ void* ipc_rcu_alloc(int size)
}
} else {
out = kmalloc(HDRLEN_KMALLOC + size, GFP_KERNEL);
+ /* the stored pointer is different from the address of
+ * the allocated block because of padding */
+ memleak_padding(out, HDRLEN_KMALLOC, size);
if (out) {
out += HDRLEN_KMALLOC;
container_of(out, struct ipc_rcu_hdr, data)->is_vmalloc = 0;
diff --git a/kernel/params.c b/kernel/params.c
index 91aea7a..b957b86 100644
--- a/kernel/params.c
+++ b/kernel/params.c
@@ -547,6 +547,7 @@ static void __init kernel_param_sysfs_se
unsigned int name_skip)
{
struct module_kobject *mk;
+ struct module_param_attrs *mp;

mk = kzalloc(sizeof(struct module_kobject), GFP_KERNEL);
BUG_ON(!mk);
@@ -556,8 +557,13 @@ static void __init kernel_param_sysfs_se
kobject_set_name(&mk->kobj, name);
kobject_register(&mk->kobj);

+ mp = param_sysfs_setup(mk, kparam, num_params, name_skip);
+ /* this structure is not freed but the pointer is
+ * lost. However, there are other pointers to its members and
+ * the object has to be kept */
+ memleak_not_leak(mp);
/* no need to keep the kobject if no parameter is exported */
- if (!param_sysfs_setup(mk, kparam, num_params, name_skip)) {
+ if (!mp) {
kobject_unregister(&mk->kobj);
kfree(mk);
}
diff --git a/mm/slab.c b/mm/slab.c
index 42c90aa..b29662e 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -3421,6 +3421,10 @@ void *__alloc_percpu(size_t size)
memset(pdata->ptrs[i], 0, size);
}

+ /* the code below changes the value of the returned pointer
+ * and kmemleak cannot find the original value during
+ * scanning. It is marked as not being a leak */
+ memleak_not_leak(pdata);
/* Catch derefs w/o wrappers */
return (void *)(~(unsigned long)pdata);

diff --git a/net/core/dev.c b/net/core/dev.c
index d95e262..3c1bcac 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -3211,6 +3211,12 @@ struct net_device *alloc_netdev(int size
dev = (struct net_device *)
(((long)p + NETDEV_ALIGN_CONST) & ~NETDEV_ALIGN_CONST);
dev->padded = (char *)dev - (char *)p;
+ /* kmemleak cannot guess the object type because the block
+ * size is different from the object size. The stored pointer
+ * is also different from the address of the allocated block
+ * because of padding */
+ memleak_padding(p, dev->padded, alloc_size - dev->padded);
+ memleak_typeid(p, struct net_device);

if (sizeof_priv)
dev->priv = netdev_priv(dev);
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 022d889..7b511e6 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -151,6 +151,9 @@ struct sk_buff *__alloc_skb(unsigned int

/* Get the HEAD */
skb = kmem_cache_alloc(cache, gfp_mask & ~__GFP_DMA);
+ /* the skbuff_fclone_cache contains objects larger than
+ * "struct sk_buff" and kmemleak cannot guess the type */
+ memleak_typeid(skb, struct sk_buff);
if (!skb)
goto out;

diff --git a/net/ipv4/netfilter/ip_conntrack_core.c b/net/ipv4/netfilter/ip_conntrack_core.c
index aa45917..7e64fe0 100644
--- a/net/ipv4/netfilter/ip_conntrack_core.c
+++ b/net/ipv4/netfilter/ip_conntrack_core.c
@@ -654,6 +654,11 @@ struct ip_conntrack *ip_conntrack_alloc(
}

conntrack = kmem_cache_alloc(ip_conntrack_cachep, GFP_ATOMIC);
+ /* tuplehash_to_ctrack doesn't pass a constant argument to
+ * container_of and therefore the conntrack->tuplehash[].list
+ * aliases are ignored */
+ memleak_container(struct ip_conntrack, tuplehash[IP_CT_DIR_ORIGINAL]);
+ memleak_container(struct ip_conntrack, tuplehash[IP_CT_DIR_REPLY]);
if (!conntrack) {
DEBUGP("Can't allocate conntrack.\n");
return ERR_PTR(-ENOMEM);
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 0834c2e..7fa6aab 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -437,6 +437,12 @@ struct Qdisc *qdisc_alloc(struct net_dev
goto errout;
sch = (struct Qdisc *) QDISC_ALIGN((unsigned long) p);
sch->padded = (char *) sch - (char *) p;
+ /* kmemleak cannot guess the object type because the block
+ * size is different from the object size. The stored pointer
+ * is also different from the address of the allocated block
+ * because of padding */
+ memleak_padding(p, sch->padded, sizeof(struct Qdisc));
+ memleak_typeid(p, struct Qdisc);

INIT_LIST_HEAD(&sch->list);
skb_queue_head_init(&sch->q);

2006-08-12 22:01:13

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 08/10] Keep the __init functions after initialization

From: Catalin Marinas <[email protected]>

This patch adds the CONFIG_DEBUG_KEEP_INIT option which preserves the
.init.text section after initialization. Memory leaks happening during this
phase can be more easily tracked.

Signed-off-by: Catalin Marinas <[email protected]>
---

include/linux/init.h | 4 ++++
lib/Kconfig.debug | 12 ++++++++++++
2 files changed, 16 insertions(+), 0 deletions(-)

diff --git a/include/linux/init.h b/include/linux/init.h
index 6667785..1590e9a 100644
--- a/include/linux/init.h
+++ b/include/linux/init.h
@@ -40,7 +40,11 @@ #include <linux/compiler.h>

/* These are for everybody (although not all archs will actually
discard it in modules) */
+#ifdef CONFIG_DEBUG_KEEP_INIT
+#define __init
+#else
#define __init __attribute__ ((__section__ (".init.text")))
+#endif
#define __initdata __attribute__ ((__section__ (".init.data")))
#define __exitdata __attribute__ ((__section__(".exit.data")))
#define __exit_call __attribute_used__ __attribute__ ((__section__ (".exitcall.exit")))
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 77ddb7d..c8ca3d6 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -185,6 +185,18 @@ config DEBUG_MEMLEAK_REPORTS_NR
reading the /sys/kernel/debug/memleak file could lead to
some soft-locks.

+config DEBUG_KEEP_INIT
+ bool "Do not free the __init functions"
+ default n
+ depends on DEBUG_MEMLEAK
+ help
+ This option moves the __init functions out of the .init.text
+ section and therefore they are no longer freed after the
+ kernel initialization. It is useful for identifying memory
+ leaks happening during the kernel or modules initialization.
+
+ If unsure, say N.
+
config DEBUG_PREEMPT
bool "Debug preemptible kernel"
depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT

2006-08-12 22:01:30

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 09/10] Simple testing for kmemleak

From: Catalin Marinas <[email protected]>

This patch only contains some very simple testing at the moment. Proper
testing will be needed.

Signed-off-by: Catalin Marinas <[email protected]>
---

lib/Kconfig.debug | 11 +++++++
mm/Makefile | 1 +
mm/memleak-test.c | 89 +++++++++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 101 insertions(+), 0 deletions(-)

diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index c8ca3d6..c2ba98d 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -197,6 +197,17 @@ config DEBUG_KEEP_INIT

If unsure, say N.

+config DEBUG_MEMLEAK_TEST
+ tristate "Test the kernel memory leak detector"
+ default n
+ depends on DEBUG_MEMLEAK
+ help
+ Say Y or M here to build the test harness for the kernel
+ memory leak detector. At the moment, this option enables a
+ module that explicitly leaks memory.
+
+ If unsure, say N.
+
config DEBUG_PREEMPT
bool "Debug preemptible kernel"
depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
diff --git a/mm/Makefile b/mm/Makefile
index edccbc0..fd46a73 100644
--- a/mm/Makefile
+++ b/mm/Makefile
@@ -24,3 +24,4 @@ obj-$(CONFIG_MEMORY_HOTPLUG) += memory_h
obj-$(CONFIG_FS_XIP) += filemap_xip.o
obj-$(CONFIG_MIGRATION) += migrate.o
obj-$(CONFIG_DEBUG_MEMLEAK) += memleak.o
+obj-$(CONFIG_DEBUG_MEMLEAK_TEST) += memleak-test.o
diff --git a/mm/memleak-test.c b/mm/memleak-test.c
new file mode 100644
index 0000000..6e603fd
--- /dev/null
+++ b/mm/memleak-test.c
@@ -0,0 +1,89 @@
+/*
+ * mm/memleak-test.c
+ *
+ * Copyright (C) 2006 ARM Limited
+ * Written by Catalin Marinas <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/list.h>
+
+#include <linux/memleak.h>
+
+struct test_node {
+ long header[25];
+ struct list_head list;
+ long footer[25];
+};
+
+static LIST_HEAD(test_list);
+
+/* Some very simple testing. This function needs to be extended for
+ * proper testing */
+static int __init memleak_test_init(void)
+{
+ struct test_node *elem;
+ int i;
+
+ printk(KERN_INFO "KMemLeak testing\n");
+
+ /* make some orphan pointers */
+ kmalloc(32, GFP_KERNEL);
+ kmalloc(32, GFP_KERNEL);
+ kmalloc(1024, GFP_KERNEL);
+ kmalloc(1024, GFP_KERNEL);
+ kmalloc(2048, GFP_KERNEL);
+ kmalloc(2048, GFP_KERNEL);
+ kmalloc(4096, GFP_KERNEL);
+ kmalloc(4096, GFP_KERNEL);
+#ifndef CONFIG_MODULES
+ kmem_cache_alloc(files_cachep, GFP_KERNEL);
+ kmem_cache_alloc(files_cachep, GFP_KERNEL);
+#endif
+ vmalloc(64);
+ vmalloc(64);
+
+ /* add elements to a list. They should only appear as orphan
+ * after the module is removed */
+ for (i = 0; i < 10; i++) {
+ elem = kmalloc(sizeof(*elem), GFP_KERNEL);
+ if (!elem)
+ return -ENOMEM;
+ memset(elem, 0, sizeof(*elem));
+ INIT_LIST_HEAD(&elem->list);
+
+ list_add_tail(&elem->list, &test_list);
+ }
+
+ return 0;
+}
+module_init(memleak_test_init);
+
+static void __exit memleak_test_exit(void)
+{
+ struct test_node *elem, *tmp;
+
+ /* remove the list elements without actually freeing the memory */
+ list_for_each_entry_safe(elem, tmp, &test_list, list)
+ list_del(&elem->list);
+}
+module_exit(memleak_test_exit);
+
+MODULE_LICENSE("GPL");

2006-08-12 22:02:00

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 10/10] Update the MAINTAINERS file for kmemleak

From: Catalin Marinas <[email protected]>

Signed-off-by: Catalin Marinas <[email protected]>
---

MAINTAINERS | 6 ++++++
1 files changed, 6 insertions(+), 0 deletions(-)

diff --git a/MAINTAINERS b/MAINTAINERS
index e3e1515..6dccfbf 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1663,6 +1663,12 @@ L: [email protected]
W: http://www.kerneljanitors.org/
S: Maintained

+KERNEL MEMORY LEAK DETECTOR
+P: Catalin Marinas
+M: [email protected]
+W: http://www.procode.org/
+S: Maintained
+
KERNEL NFSD
P: Neil Brown
M: [email protected]

2006-08-12 22:02:01

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 04/10] Modules support for kmemleak

From: Catalin Marinas <[email protected]>

This patch handles the kmemleak operations needed for modules loading so
that memory allocations from inside a module are properly tracked.

Signed-off-by: Catalin Marinas <[email protected]>
---

kernel/module.c | 41 +++++++++++++++++++++++++++++++++++++++++
1 files changed, 41 insertions(+), 0 deletions(-)

diff --git a/kernel/module.c b/kernel/module.c
index 2a19cd4..a7f8c6d 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -1481,6 +1481,11 @@ static struct module *load_module(void _
unsigned int unusedcrcindex;
unsigned int unusedgplindex;
unsigned int unusedgplcrcindex;
+#ifdef CONFIG_DEBUG_MEMLEAK
+ unsigned int dataindex;
+ unsigned int bssindex;
+ unsigned int mloffindex;
+#endif
struct module *mod;
long err = 0;
void *percpu = NULL, *ptr = NULL; /* Stops spurious gcc warning */
@@ -1577,6 +1582,11 @@ #endif
#ifdef ARCH_UNWIND_SECTION_NAME
unwindex = find_sec(hdr, sechdrs, secstrings, ARCH_UNWIND_SECTION_NAME);
#endif
+#ifdef CONFIG_DEBUG_MEMLEAK
+ dataindex = find_sec(hdr, sechdrs, secstrings, ".data");
+ bssindex = find_sec(hdr, sechdrs, secstrings, ".bss");
+ mloffindex = find_sec(hdr, sechdrs, secstrings, ".init.memleak_offsets");
+#endif

/* Don't keep modinfo section */
sechdrs[infoindex].sh_flags &= ~(unsigned long)SHF_ALLOC;
@@ -1646,6 +1656,10 @@ #endif

/* Do the allocs. */
ptr = module_alloc(mod->core_size);
+ /* the pointer to this block is stored in the module structure
+ * which is inside the block. Just mark it as not being a
+ * leak */
+ memleak_not_leak(ptr);
if (!ptr) {
err = -ENOMEM;
goto free_percpu;
@@ -1654,6 +1668,11 @@ #endif
mod->module_core = ptr;

ptr = module_alloc(mod->init_size);
+ /* the pointer to this block is stored in the module structure
+ * which is inside the block. This block doesn't need to be
+ * scanned as it contains data and code that will be freed
+ * after the module is initialized */
+ memleak_ignore(ptr);
if (!ptr && mod->init_size) {
err = -ENOMEM;
goto free_core;
@@ -1685,6 +1704,28 @@ #endif
/* Module has been moved. */
mod = (void *)sechdrs[modindex].sh_addr;

+#ifdef CONFIG_DEBUG_MEMLEAK
+ if (mloffindex)
+ memleak_insert_aliases((void *)sechdrs[mloffindex].sh_addr,
+ (void *)sechdrs[mloffindex].sh_addr
+ + sechdrs[mloffindex].sh_size);
+
+ /* only scan the sections containing data */
+ memleak_scan_area(mod->module_core,
+ (unsigned long)mod - (unsigned long)mod->module_core,
+ sizeof(struct module));
+ if (dataindex)
+ memleak_scan_area(mod->module_core,
+ sechdrs[dataindex].sh_addr
+ - (unsigned long)mod->module_core,
+ sechdrs[dataindex].sh_size);
+ if (bssindex)
+ memleak_scan_area(mod->module_core,
+ sechdrs[bssindex].sh_addr
+ - (unsigned long)mod->module_core,
+ sechdrs[bssindex].sh_size);
+#endif
+
/* Now we've moved module, initialize linked lists, etc. */
module_unload_init(mod);

2006-08-12 22:02:45

by Catalin Marinas

[permalink] [raw]
Subject: [PATCH 2.6.18-rc4 06/10] Add kmemleak support for ARM

From: Catalin Marinas <[email protected]>

This patch modifies the vmlinux.lds.S script and adds the backtrace support
for ARM to be used with kmemleak.

Signed-off-by: Catalin Marinas <[email protected]>
---

arch/arm/kernel/vmlinux.lds.S | 7 +++++++
include/asm-arm/processor.h | 12 ++++++++++++
2 files changed, 19 insertions(+), 0 deletions(-)

diff --git a/arch/arm/kernel/vmlinux.lds.S b/arch/arm/kernel/vmlinux.lds.S
index 3ca574e..59976b8 100644
--- a/arch/arm/kernel/vmlinux.lds.S
+++ b/arch/arm/kernel/vmlinux.lds.S
@@ -67,6 +67,11 @@ #endif
__per_cpu_start = .;
*(.data.percpu)
__per_cpu_end = .;
+#ifdef CONFIG_DEBUG_MEMLEAK
+ __memleak_offsets_start = .;
+ *(.init.memleak_offsets)
+ __memleak_offsets_end = .;
+#endif
#ifndef CONFIG_XIP_KERNEL
__init_begin = _stext;
*(.init.data)
@@ -115,6 +120,7 @@ #endif

.data : AT(__data_loc) {
__data_start = .; /* address in memory */
+ _sdata = .;

/*
* first, the init task union, aligned
@@ -165,6 +171,7 @@ #endif
__bss_start = .; /* BSS */
*(.bss)
*(COMMON)
+ __bss_stop = .;
_end = .;
}
/* Stabs debugging sections. */
diff --git a/include/asm-arm/processor.h b/include/asm-arm/processor.h
index 04f4d34..34a3bb3 100644
--- a/include/asm-arm/processor.h
+++ b/include/asm-arm/processor.h
@@ -121,6 +121,18 @@ #define spin_lock_prefetch(x) do { } whi

#endif

+#ifdef CONFIG_FRAME_POINTER
+static inline unsigned long arch_call_address(void *frame)
+{
+ return *(unsigned long *)(frame - 4) - 4;
+}
+
+static inline void *arch_prev_frame(void *frame)
+{
+ return *(void **)(frame - 12);
+}
+#endif
+
#endif

#endif /* __ASM_ARM_PROCESSOR_H */

2006-08-13 11:59:47

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

Hi Catalin,

On 12/08/06, Catalin Marinas <[email protected]> wrote:
> This is a new version (0.9) of the kernel memory leak detector. See
> the Documentation/kmemleak.txt file for a more detailed
> description. The patches are downloadable from (the whole patch or the
> broken-out series):
>
> http://homepage.ntlworld.com/cmarinas/kmemleak/patch-2.6.18-rc4-kmemleak-0.9.bz2
> http://homepage.ntlworld.com/cmarinas/kmemleak/broken-out/patches-kmemleak-0.9.tar.bz2

Can you look at this?

=======================================================
[ INFO: possible circular locking dependency detected ]
-------------------------------------------------------
events/0/8 is trying to acquire lock:
(old_style_spin_init){++..}, at: [<c017674f>] memleak_free+0x95/0x157

but task is already holding lock:
(&parent->list_lock){++..}, at: [<c0174f29>] drain_array+0x49/0xad

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #1 (&parent->list_lock){++..}:
[<c0140cc7>] check_prevs_add+0x4d/0xaf
[<c01423c1>] __lock_acquire+0x7b1/0x814
[<c01429bc>] lock_acquire+0x5e/0x7e
[<c02f9f7a>] _spin_lock+0x23/0x2f
[<c0174058>] cache_alloc_refill+0x76/0x1d2
[<c0174559>] kmem_cache_alloc+0x73/0xce
[<c01f0c8a>] radix_tree_node_alloc+0x1a/0x51
[<c01f0e3f>] radix_tree_insert+0x51/0xfb
[<c01761f6>] insert_alias+0x85/0xe8
[<c01762a4>] memleak_insert_aliases+0x4b/0xa6
[<c01773f7>] memleak_init+0x44/0xf5
[<c0100ab0>] start_kernel+0x17e/0x1f9
[<c0100210>] 0xc0100210
-> #0 (old_style_spin_init){++..}:
[<c0140cc7>] check_prevs_add+0x4d/0xaf
[<c01423c1>] __lock_acquire+0x7b1/0x814
[<c01429bc>] lock_acquire+0x5e/0x7e
[<c02f9f7a>] _spin_lock+0x23/0x2f
[<c017674f>] memleak_free+0x95/0x157
[<c0174a74>] kmem_cache_free+0x62/0xbc
[<c0172fc8>] slab_destroy+0x48/0x4d
[<c01743b8>] free_block+0xc9/0x101
[<c0174f65>] drain_array+0x85/0xad
[<c017500d>] cache_reap+0x80/0xfe
[<c01394dd>] run_workqueue+0x88/0xc4
[<c0139617>] worker_thread+0xfe/0x131
[<c013c6e1>] kthread+0x82/0xaa
[<c01044c9>] kernel_thread_helper+0x5/0xb

other info that might help us debug this:

2 locks held by events/0/8:
#0: (cache_chain_mutex){--..}, at: [<c0174f9d>] cache_reap+0x10/0xfe
#1: (&parent->list_lock){++..}, at: [<c0174f29>] drain_array+0x49/0xad
stack backtrace:
[<c0106e59>] show_trace_log_lvl+0x58/0x14c
[<c0106f5a>] show_trace+0xd/0xf
[<c010702c>] dump_stack+0x17/0x19
[<c01405c0>] print_circular_bug_tail+0x59/0x62
[<c0140af1>] check_prev_add+0x2b/0x1b4
[<c0140cc7>] check_prevs_add+0x4d/0xaf
[<c01423c1>] __lock_acquire+0x7b1/0x814
[<c01429bc>] lock_acquire+0x5e/0x7e
[<c02f9f7a>] _spin_lock+0x23/0x2f
[<c017674f>] memleak_free+0x95/0x157
[<c0174a74>] kmem_cache_free+0x62/0xbc
[<c0172fc8>] slab_destroy+0x48/0x4d
[<c01743b8>] free_block+0xc9/0x101
[<c0174f65>] drain_array+0x85/0xad
[<c017500d>] cache_reap+0x80/0xfe
[<c01394dd>] run_workqueue+0x88/0xc4
[<c0139617>] worker_thread+0xfe/0x131
[<c013c6e1>] kthread+0x82/0xaa
[<c01044c9>] kernel_thread_helper+0x5/0xb
DWARF2 unwinder stuck at kernel_thread_helper+0x5/0xb
Leftover inexact backtrace:
[<c0106f5a>] show_trace+0xd/0xf
[<c010702c>] dump_stack+0x17/0x19
[<c01405c0>] print_circular_bug_tail+0x59/0x62
[<c0140af1>] check_prev_add+0x2b/0x1b4
[<c0140cc7>] check_prevs_add+0x4d/0xaf
[<c01423c1>] __lock_acquire+0x7b1/0x814
[<c01429bc>] lock_acquire+0x5e/0x7e
[<c02f9f7a>] _spin_lock+0x23/0x2f
[<c017674f>] memleak_free+0x95/0x157
[<c0174a74>] kmem_cache_free+0x62/0xbc
[<c0172fc8>] slab_destroy+0x48/0x4d
[<c01743b8>] free_block+0xc9/0x101
[<c0174f65>] drain_array+0x85/0xad
[<c017500d>] cache_reap+0x80/0xfe
[<c01394dd>] run_workqueue+0x88/0xc4
[<c0139617>] worker_thread+0xfe/0x131
[<c013c6e1>] kthread+0x82/0xaa
[<c01044c9>] kernel_thread_helper+0x5/0xb

0xc017674f is in memleak_free (/usr/src/linux-work3/mm/memleak.c:479).
474 /* Remove a pointer and its aliases from the pointer radix tree */
475 static inline void delete_pointer(unsigned long ptr)
476 {
477 struct memleak_pointer *pointer;
478
479 pointer = radix_tree_delete(&pointer_tree, ptr);
480 if (!pointer) {
481 dump_stack();
482 printk(KERN_WARNING "kmemleak: freeing unknown
pointer value 0x%08lx\n", ptr);
483 return;

0xc0174f29 is in drain_array (/usr/src/linux-work3/mm/slab.c:3739).
3734 return;
3735 if (ac->touched && !force) {
3736 ac->touched = 0;
3737 } else {
3738 spin_lock_irq(&l3->list_lock);
3739 if (ac->avail) {
3740 tofree = force ? ac->avail :
(ac->limit + 4) / 5;
3741 if (tofree > ac->avail)
3742 tofree = (ac->avail + 1) / 2;
3743 free_block(cachep, ac->entry, tofree, node);

0xc0174f9d is in cache_reap (/usr/src/linux-work3/mm/slab.c:3770).
3765 {
3766 struct kmem_cache *searchp;
3767 struct kmem_list3 *l3;
3768 int node = numa_node_id();
3769
3770 if (!mutex_trylock(&cache_chain_mutex)) {
3771 /* Give up. Setup the next iteration. */
3772 schedule_delayed_work(&__get_cpu_var(reap_work),
3773 REAPTIMEOUT_CPUC);
3774 return;

config file and dmesg http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/

> --
> Catalin

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-13 14:13:49

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

Hi Michal,

On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> Can you look at this?
>
=======================================================
> [ INFO: possible circular locking dependency detected ]

Thanks for this. Did you get this with the previous version of
kmemleak (0.8) or you just tried it now.

I'll have a look.

--
Catalin

2006-08-13 14:26:24

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 13/08/06, Catalin Marinas <[email protected]> wrote:
> Hi Michal,
>
> On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> > Can you look at this?
> >
> =======================================================
> > [ INFO: possible circular locking dependency detected ]
>
> Thanks for this. Did you get this with the previous version of
> kmemleak (0.8) or you just tried it now.

It's kmemleak 0.9 issue. I have tested kmemleak 0.8 on 2.6.18-rc1and
2.6.18-rc2. I haven't seen this before.

>
> I'll have a look.
>
> --
> Catalin
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-14 15:34:44

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> Can you look at this?
> =======================================================
> [ INFO: possible circular locking dependency detected ]
> -------------------------------------------------------
> events/0/8 is trying to acquire lock:
> (old_style_spin_init){++..}, at: [<c017674f>] memleak_free+0x95/0x157
>
> but task is already holding lock:
> (&parent->list_lock){++..}, at: [<c0174f29>] drain_array+0x49/0xad
>
> which lock already depends on the new lock.
>
>
> the existing dependency chain (in reverse order) is:
>
> -> #1 (&parent->list_lock){++..}:
> [<c0140cc7>] check_prevs_add+0x4d/0xaf
> [<c01423c1>] __lock_acquire+0x7b1/0x814
> [<c01429bc>] lock_acquire+0x5e/0x7e
> [<c02f9f7a>] _spin_lock+0x23/0x2f
> [<c0174058>] cache_alloc_refill+0x76/0x1d2
> [<c0174559>] kmem_cache_alloc+0x73/0xce
> [<c01f0c8a>] radix_tree_node_alloc+0x1a/0x51
> [<c01f0e3f>] radix_tree_insert+0x51/0xfb
> [<c01761f6>] insert_alias+0x85/0xe8
> [<c01762a4>] memleak_insert_aliases+0x4b/0xa6
> [<c01773f7>] memleak_init+0x44/0xf5
> [<c0100ab0>] start_kernel+0x17e/0x1f9
> [<c0100210>] 0xc0100210
> -> #0 (old_style_spin_init){++..}:
> [<c0140cc7>] check_prevs_add+0x4d/0xaf
> [<c01423c1>] __lock_acquire+0x7b1/0x814
> [<c01429bc>] lock_acquire+0x5e/0x7e
> [<c02f9f7a>] _spin_lock+0x23/0x2f
> [<c017674f>] memleak_free+0x95/0x157
> [<c0174a74>] kmem_cache_free+0x62/0xbc
> [<c0172fc8>] slab_destroy+0x48/0x4d
> [<c01743b8>] free_block+0xc9/0x101
> [<c0174f65>] drain_array+0x85/0xad
> [<c017500d>] cache_reap+0x80/0xfe
> [<c01394dd>] run_workqueue+0x88/0xc4
> [<c0139617>] worker_thread+0xfe/0x131
> [<c013c6e1>] kthread+0x82/0xaa
> [<c01044c9>] kernel_thread_helper+0x5/0xb

I don't think I fully understand the slab locking, maybe some other
kernel guys could help with this, but lockdep is probably right (the
lock could happen on SMP systems).

It looks like lockdep complains that memleak_lock is acquired while
list_lock is held and there is a possibility that (on a different CPU)
memleak_lock was acquired first, followed by list_lock acquiring and
therefore a deadlock.

The kmemleak+slab locking is a bit complicated because memleak itself
needs to allocate memory and avoid recursive calls to it (the
pointer_cache and the radix_tree allocations). The kmemleak-related
allocations are not tracked by kmemleak.

I see the following solutions:

1. acquire the memleak_lock at the beginning of an alloc/free function
and release it when finished while allowing recursive/nested calls
(and only call the memleak_* functions during the outermost lock).
This would mean ignoring the off-slab management allocations as these
would lead to deadlock because of the recursive call into kmemleak.
This locking should be placed around cache_reap() as well (actually,
around all the entry points in the mm/slab.c file).

2. do an independent (simple) memory management in kmemleak and
probably replace radix_tree with prio_tree as the latter doesn't seem
to require allocations.

The first option is simple to implement but it has the disadvantage of
serializing the slab calls on SMP and also not tracking the mm/slab.c
allocations. The second one would provide full coverage of the kernel
slab allocations but it is probably more difficult to implement.

Any thoughts/suggestions on this?

Thanks.

--
Catalin

2006-08-15 15:25:40

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 14/08/06, Catalin Marinas <[email protected]> wrote:
> The kmemleak+slab locking is a bit complicated because memleak itself
> needs to allocate memory and avoid recursive calls to it (the
> pointer_cache and the radix_tree allocations). The kmemleak-related
> allocations are not tracked by kmemleak.
>
> I see the following solutions:
>
> 1. acquire the memleak_lock at the beginning of an alloc/free function
> and release it when finished while allowing recursive/nested calls
> (and only call the memleak_* functions during the outermost lock).
> This would mean ignoring the off-slab management allocations as these
> would lead to deadlock because of the recursive call into kmemleak.
> This locking should be placed around cache_reap() as well (actually,
> around all the entry points in the mm/slab.c file).

This would actually work because the slab allocation functions may sleep.

The simple solution would be to eliminate the memory allocations when
memleak_lock is held, maybe by replacing the radix tree with a prio
search tree (not sure what the performance impact will be). I'll try
to go this route and see how it works.

--
Catalin

2006-08-15 15:26:37

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 15/08/06, Catalin Marinas <[email protected]> wrote:
> On 14/08/06, Catalin Marinas <[email protected]> wrote:
> > The kmemleak+slab locking is a bit complicated because memleak itself
> > needs to allocate memory and avoid recursive calls to it (the
> > pointer_cache and the radix_tree allocations). The kmemleak-related
> > allocations are not tracked by kmemleak.
> >
> > I see the following solutions:
> >
> > 1. acquire the memleak_lock at the beginning of an alloc/free function
> > and release it when finished while allowing recursive/nested calls
> > (and only call the memleak_* functions during the outermost lock).
> > This would mean ignoring the off-slab management allocations as these
> > would lead to deadlock because of the recursive call into kmemleak.
> > This locking should be placed around cache_reap() as well (actually,
> > around all the entry points in the mm/slab.c file).
>
> This would actually work because the slab allocation functions may sleep.

I meant "would *not*" above.

--
Catalin

2006-08-17 13:02:13

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> It's kmemleak 0.9 issue. I have tested kmemleak 0.8 on 2.6.18-rc1and
> 2.6.18-rc2. I haven't seen this before.

it looks like it was caused by commit
fc818301a8a39fedd7f0a71f878f29130c72193d where free_block() now calls
slab_destroy() with l3->list_lock held.

The prio_tree use (which doesn't alloc memory) instead of the
radix_tree is about 4 times slower when scanning the memory and I
don't think I'll use it.

It leaves me with the options of either implementing my own memory
allocator based on pages (including a simple hash table instead of
radix tree) or fix the locking in kmemleak so that memory allocations
happen without memleak_lock held. The latter is a bit complicated as
well since any slab allocation causes a re-entrance into kmemleak.

Any other suggestions?

Thanks.

--
Catalin

2006-08-17 14:45:09

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

Hi Catalin,

On 17/08/06, Catalin Marinas <[email protected]> wrote:
> On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> > It's kmemleak 0.9 issue. I have tested kmemleak 0.8 on 2.6.18-rc1and
> > 2.6.18-rc2. I haven't seen this before.
>
> it looks like it was caused by commit
> fc818301a8a39fedd7f0a71f878f29130c72193d where free_block() now calls
> slab_destroy() with l3->list_lock held.

I'll revert this commit.

>
> The prio_tree use (which doesn't alloc memory) instead of the
> radix_tree is about 4 times slower when scanning the memory and I
> don't think I'll use it.
>
> It leaves me with the options of either implementing my own memory
> allocator based on pages (including a simple hash table instead of
> radix tree) or fix the locking in kmemleak so that memory allocations
> happen without memleak_lock held. The latter is a bit complicated as
> well since any slab allocation causes a re-entrance into kmemleak.
>
> Any other suggestions?

Please talk with Christoph Lameter, he is working on Modular Slab.
http://www.ussg.iu.edu/hypermail/linux/kernel/0608.1/0951.html
http://www.ussg.iu.edu/hypermail/linux/kernel/0608.2/0030.html
Maybe he can help with this problem.

>
> Thanks.
>
> --
> Catalin
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-17 14:48:36

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 17/08/06, Michal Piotrowski <[email protected]> wrote:
> Hi Catalin,
>
> On 17/08/06, Catalin Marinas <[email protected]> wrote:
> > On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> > > It's kmemleak 0.9 issue. I have tested kmemleak 0.8 on 2.6.18-rc1and
> > > 2.6.18-rc2. I haven't seen this before.
> >
> > it looks like it was caused by commit
> > fc818301a8a39fedd7f0a71f878f29130c72193d where free_block() now calls
> > slab_destroy() with l3->list_lock held.
>
> I'll revert this commit.
>
> >
> > The prio_tree use (which doesn't alloc memory) instead of the
> > radix_tree is about 4 times slower when scanning the memory and I
> > don't think I'll use it.
> >
> > It leaves me with the options of either implementing my own memory
> > allocator based on pages

[MODSLAB 7/7] A slab allocator: Page Slab allocator
"The page slab is a specialized slab allocator that can only handle
page order size object. It directly uses the page allocator to
track the objects and can therefore avoid the overhead of the
slabifier."
http://www.ussg.iu.edu/hypermail/linux/kernel/0608.1/3023.html

>(including a simple hash table instead of
> > radix tree) or fix the locking in kmemleak so that memory allocations
> > happen without memleak_lock held. The latter is a bit complicated as
> > well since any slab allocation causes a re-entrance into kmemleak.
> >
> > Any other suggestions?
>
> Please talk with Christoph Lameter, he is working on Modular Slab.
> http://www.ussg.iu.edu/hypermail/linux/kernel/0608.1/0951.html
> http://www.ussg.iu.edu/hypermail/linux/kernel/0608.2/0030.html
> Maybe he can help with this problem.
>
> >
> > Thanks.
> >
> > --
> > Catalin
> >

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-17 15:01:49

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 17/08/06, Michal Piotrowski <[email protected]> wrote:
> On 17/08/06, Catalin Marinas <[email protected]> wrote:
> > On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> > > It's kmemleak 0.9 issue. I have tested kmemleak 0.8 on 2.6.18-rc1and
> > > 2.6.18-rc2. I haven't seen this before.
> >
> > it looks like it was caused by commit
> > fc818301a8a39fedd7f0a71f878f29130c72193d where free_block() now calls
> > slab_destroy() with l3->list_lock held.
>
> I'll revert this commit.

I'm not sure it's a good idea, it might have other implications in
slab.c. I better fix kmemleak (I think currently you could get a
deadlock only on SMP).

> Please talk with Christoph Lameter, he is working on Modular Slab.
> http://www.ussg.iu.edu/hypermail/linux/kernel/0608.1/0951.html
> http://www.ussg.iu.edu/hypermail/linux/kernel/0608.2/0030.html
> Maybe he can help with this problem.

I haven't looked at these patches in detail but they look like making
the slab allocator cleaner.

Anyway, I still need to revisit the locking in kmemleak and not rely
on future changes to slab.c. At the moment I think I can avoid any
kmemleak locks when allocating memory (by using radix_tree_preload
with the radix trees). If this still fails, I'll think about writing
my own, very simple, memory allocator and avoid the re-entrance
problem.

Thanks.

--
Catalin

2006-08-17 15:05:01

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 17/08/06, Michal Piotrowski <[email protected]> wrote:
> > On 17/08/06, Catalin Marinas <[email protected]> wrote:
> > > It leaves me with the options of either implementing my own memory
> > > allocator based on pages
>
> [MODSLAB 7/7] A slab allocator: Page Slab allocator
> "The page slab is a specialized slab allocator that can only handle
> page order size object. It directly uses the page allocator to
> track the objects and can therefore avoid the overhead of the
> slabifier."
> http://www.ussg.iu.edu/hypermail/linux/kernel/0608.1/3023.html

But this one allocates page-order size objects. I usually have plenty
of small objects in kmemleak.

--
Catalin

2006-08-17 21:58:29

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 17/08/06, Catalin Marinas <[email protected]> wrote:
> On 17/08/06, Michal Piotrowski <[email protected]> wrote:
> > On 17/08/06, Catalin Marinas <[email protected]> wrote:
> > > On 13/08/06, Michal Piotrowski <[email protected]> wrote:
> > > > It's kmemleak 0.9 issue. I have tested kmemleak 0.8 on 2.6.18-rc1and
> > > > 2.6.18-rc2. I haven't seen this before.
> > >
> > > it looks like it was caused by commit
> > > fc818301a8a39fedd7f0a71f878f29130c72193d where free_block() now calls
> > > slab_destroy() with l3->list_lock held.
> >
> > I'll revert this commit.
>
> I'm not sure it's a good idea, it might have other implications in
> slab.c. I better fix kmemleak (I think currently you could get a
> deadlock only on SMP).

I attach a patch for kmemleak-0.9 which seems to work for me. It has a
new locking mechanism in that memory allocations are performed without
memleak_lock held. Let me know if you still get errors (I haven't
fully tested it yet).

Thanks.

--
Catalin


Attachments:
(No filename) (1.01 kB)
new-locking.patch (25.30 kB)
Download all attachments

2006-08-18 12:28:15

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

Hi Catalin,

On 17/08/06, Catalin Marinas <[email protected]> wrote:
> On 17/08/06, Catalin Marinas <[email protected]> wrote:
> > I'm not sure it's a good idea, it might have other implications in
> > slab.c. I better fix kmemleak (I think currently you could get a
> > deadlock only on SMP).
>
> I attach a patch for kmemleak-0.9 which seems to work for me. It has a
> new locking mechanism in that memory allocations are performed without
> memleak_lock held. Let me know if you still get errors (I haven't
> fully tested it yet).

Something doesn't work. It appears while udev start.

[<c0106cdc>] show_trace+0xd/0xf
[<c0106dae>] dump_stack+0x17/0x19
[<c016e9a0>] __kmalloc+0xf6/0x10a
[<c019a4bf>] load_elf_binary+0x5b/0xb46
[<c017d77a>] search_binary_handler+0xfc/0x2d4
[<c017daa5>] do_execve+0x153/0x1f0
[<c0104d5f>] sys_execve+0x29/0x7a
[<c0105f11>] sysenter_past_esp+0x5b/0x79
kmemleak: pointer 0xf5a39ae0:
trace:
c016e767: <kmemleak_cache_alloc>
c0127274: <dup_mm>
c0127427: <copy_mm>
c0128227: <copy_process>
c012893: <do_fork>
c0104d19: <sys_clone>

Kernel panic - not syncing: kmemleak: resizing pointer by alias 0xf5ad4a04

BUG: warning at
/usr/src/linux-work8/arch/i386/kernel/smp.c:547/smp_call_function()
[<c0106cdc>] show_trace+0xd/0xf
[<c0106dae>] dump_stack+0x17/0x19
[<c0116ac4>] smp_call_function+0x52/0xf9
[<c0116bad>] smp_send_stop+0x16/0x20
[<c0170c35>] panic+0x50/0xdf
[<c016e9a0>] __kmalloc+0xf6/0x10a
[<c019a4bf>] load_elf_binary+0x5b/0xb46
[<c017d77a>] search_binary_handler+0xfc/0x2d4
[<c017daa5>] do_execve+0x153/0x1f0
[<c0104d5f>] sys_execve+0x29/0x7a
[<c0105f11>] sysenter_past_esp+0x5b/0x79

http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/kml-config2

>
> Thanks.
>
> --
> Catalin
>
>
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-18 13:27:46

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> Something doesn't work. It appears while udev start.

Could you try the attached patch? It has some improvements from the
yesterday's one and also prints extra information when that error
happens.

Thanks.

--
Catalin


Attachments:
(No filename) (293.00 B)
new-locking.patch (26.79 kB)
Download all attachments

2006-08-18 13:56:00

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Catalin Marinas <[email protected]> wrote:
> On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > Something doesn't work. It appears while udev start.
>
> Could you try the attached patch? It has some improvements from the
> yesterday's one and also prints extra information when that error
> happens.

Problem fixed, thank. I'll do some tests.

>
> Thanks.
>
> --
> Catalin
>
>
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-18 14:15:48

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> On 18/08/06, Catalin Marinas <[email protected]> wrote:
> > On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > > Something doesn't work. It appears while udev start.
> >
> > Could you try the attached patch? It has some improvements from the
> > yesterday's one and also prints extra information when that error
> > happens.
>
> Problem fixed, thank. I'll do some tests.

Lockdep still detects this bug

=======================================================
[ INFO: possible circular locking dependency detected ]
-------------------------------------------------------
events/0/8 is trying to acquire lock:
(memleak_lock){++..}, at: [<c0176a11>] memleak_free+0x9e/0x16f

but task is already holding lock:
(&parent->list_lock){++..}, at: [<c0174f29>] drain_array+0x49/0xad

which lock already depends on the new lock.
etc.

>
> >
> > Thanks.
> >
> > --
> > Catalin

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-18 15:46:31

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> Lockdep still detects this bug
>
> =======================================================
> [ INFO: possible circular locking dependency detected ]
> -------------------------------------------------------
> events/0/8 is trying to acquire lock:
> (memleak_lock){++..}, at: [<c0176a11>] memleak_free+0x9e/0x16f
>
> but task is already holding lock:
> (&parent->list_lock){++..}, at: [<c0174f29>] drain_array+0x49/0xad
>
> which lock already depends on the new lock.
> etc.

Do you have the complete dmesg output? There should be a different
path which I missed (I didn't get this yestarday but I haven't tried
with the latest patch I sent to you today).

--
Catalin

2006-08-18 16:42:53

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Catalin Marinas <[email protected]> wrote:
> On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > Lockdep still detects this bug
> >
> > =======================================================
> > [ INFO: possible circular locking dependency detected ]
> > -------------------------------------------------------
> > events/0/8 is trying to acquire lock:
> > (memleak_lock){++..}, at: [<c0176a11>] memleak_free+0x9e/0x16f
> >
> > but task is already holding lock:
> > (&parent->list_lock){++..}, at: [<c0174f29>] drain_array+0x49/0xad
> >
> > which lock already depends on the new lock.
> > etc.
>
> Do you have the complete dmesg output? There should be a different
> path which I missed (I didn't get this yestarday but I haven't tried
> with the latest patch I sent to you today).

Here is dmesg output
http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/kml-dmesg2

>
> --
> Catalin
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-18 16:57:29

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> On 18/08/06, Catalin Marinas <[email protected]> wrote:
> > On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > > Lockdep still detects this bug
> >
> > Do you have the complete dmesg output? There should be a different
> > path which I missed (I didn't get this yestarday but I haven't tried
> > with the latest patch I sent to you today).
>
> Here is dmesg output
> http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/kml-dmesg2

Thanks. It is strange - I was under the impression that calling
radix_tree_preload() outside the memleak_lock holding would actually
prevent radix_tree_insert() from allocating a node (and therefore call
kmem_cache_alloc and acquire the list_lock). I'll look at the
radix_tree code this weekend. I'm looking at implementing some kind of
RCU mechanism in kmemleak to avoid future problems (might need changed
to radix_tree as well).

--
Catalin

2006-08-18 21:38:32

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Catalin Marinas <[email protected]> wrote:
[snip]
> to radix_tree as well).

cat /sys/kernel/debug/memleak + LTP + patching repo = oops?

BUG: unable to handle kernel paging request at virtual address 6b6b6b6b
printing eip:
c0177468
*pde = 00000000
Oops: 0000 [#1]
PREEMPT SMP
Modules linked in: jfs nls_base xfs reiserfs ext2 loop ipv6 w83627hf
hwmon_vid hwmon i2c_isa af_packet ip_conntrack_netbios_
ns ipt_REJECT xt_state ip_conntrack nfnetlink xt_tcpudp iptable_filter
ip_tables x_tables cpufreq_userspace p4_clockmod spee
dstep_lib binfmt_misc thermal processor fan container evdev
snd_intel8x0 snd_ac97_codec snd_ac97_bus snd_seq_dummy snd_seq_o
ss snd_seq_midi_event snd_seq snd_seq_device snd_pcm_oss sk98lin
snd_mixer_oss snd_pcm skge snd_timer snd i2c_i801 intel_agp
soundcore snd_page_alloc ide_cd agpgart cdrom rtc unix
CPU: 0
EIP: 0060:[<c0177468>] Not tainted VLI
EFLAGS: 00010097 (2.6.18-rc4 #174)
EIP is at memleak_scan+0x58/0x54e
eax: 00000000 ebx: 00000000 ecx: 00000000 edx: 6b6b6b6b
esi: 6b6b6b67 edi: 00000000 ebp: de1aaf28 esp: de1aaf10
ds: 007b es: 007b ss: 0068
Process cat (pid: 6409, ti=de1aa000 task=e23ee1b0 task.ti=de1aa000)
Stack: 00000000 00000282 00000000 00000000 00000000 c035e220 de1aaf3c c017797e
00001000 e1ae03d8 c035e220 de1aaf74 c0196654 00000000 00000000 00001000
0804f000 f08e0c9c c0120ac5 00000001 00000000 00000000 00001000 f08e0c9c
Call Trace:
[<c017797e>] memleak_seq_start+0x20/0xaa
[<c0196654>] seq_read+0xc8/0x246
[<c017988b>] vfs_read+0xa9/0x151
[<c0179bab>] sys_read+0x3b/0x60
[<c0105f3d>] sysenter_past_esp+0x56/0x8d
[<b7f7c410>] 0xb7f7c410
[<c0106fe3>] show_stack_log_lvl+0x87/0x8f
[<c0107148>] show_registers+0x11a/0x183
[<c0107334>] die+0x10f/0x1da
[<c0120c69>] do_page_fault+0x3ce/0x4a4
[<c0106b41>] error_code+0x39/0x40
[<c017797e>] memleak_seq_start+0x20/0xaa
[<c0196654>] seq_read+0xc8/0x246
[<c017988b>] vfs_read+0xa9/0x151
[<c0179bab>] sys_read+0x3b/0x60
[<c0105f3d>] sysenter_past_esp+0x56/0x8d
Code: 19 c0 85 c0 74 08 0f 0b f3 04 fe bd 31 c0 b8 01 00 00 00 e8 cb
cd fa ff b8 a0 e1 35 c0 e8 e8 32 18 00 8b 15 8c e1 35 c
0 8d 72 fc <8b> 46 04 0f 18 00 90 81 fa 8c e1 35 c0 0f 84 0a 03 00 00 8b 06
EIP: [<c0177468>] memleak_scan+0x58/0x54e SS:ESP 0068:de1aaf10
<6>note: cat[6409] exited with preempt_count 3
BUG: sleeping function called from invalid context at
/usr/src/linux-work8/mm/slab.c:2904
in_atomic():1, irqs_disabled():1
[<c0106e59>] show_trace_log_lvl+0x58/0x14c
[<c0106f5a>] show_trace+0xd/0xf
[<c010702c>] dump_stack+0x17/0x19
[<c0126dc6>] __might_sleep+0x92/0x9a
[<c01745dc>] kmem_cache_zalloc+0x28/0xf5
[<c0159eee>] taskstats_exit_alloc+0x30/0x6e
[<c012c844>] do_exit+0x10e/0x46a
[<c01073f7>] die+0x1d2/0x1da
[<c0120c69>] do_page_fault+0x3ce/0x4a4
[<c0106b41>] error_code+0x39/0x40
[<c0177468>] memleak_scan+0x58/0x54e
[<c017797e>] memleak_seq_start+0x20/0xaa
[<c0196654>] seq_read+0xc8/0x246
[<c017988b>] vfs_read+0xa9/0x151
[<c0179bab>] sys_read+0x3b/0x60
[<c0105f3d>] sysenter_past_esp+0x56/0x8d
[<b7f7c410>] 0xb7f7c410
[<c0106f5a>] show_trace+0xd/0xf
[<c010702c>] dump_stack+0x17/0x19
[<c0126dc6>] __might_sleep+0x92/0x9a
[<c01745dc>] kmem_cache_zalloc+0x28/0xf5
[<c0159eee>] taskstats_exit_alloc+0x30/0x6e
[<c012c844>] do_exit+0x10e/0x46a
[<c01073f7>] die+0x1d2/0x1da
[<c0120c69>] do_page_fault+0x3ce/0x4a4
[<c0106b41>] error_code+0x39/0x40
[<c017797e>] memleak_seq_start+0x20/0xaa
[<c0196654>] seq_read+0xc8/0x246
[<c017988b>] vfs_read+0xa9/0x151
[<c0179bab>] sys_read+0x3b/0x60
[<c0105f3d>] sysenter_past_esp+0x56/0x8d
BUG: spinlock lockup on CPU#1, growfiles/6257, c035e1a0
[<c0106e59>] show_trace_log_lvl+0x58/0x14c
[<c0106f5a>] show_trace+0xd/0xf
[<c010702c>] dump_stack+0x17/0x19
[<c01fedbd>] __spin_lock_debug+0x86/0x93
[<c01fee28>] _raw_spin_lock+0x5e/0x73
[<c02fa771>] _spin_lock+0x2a/0x2f
[<c0176877>] memleak_alloc+0x17b/0x277
[<c01745a6>] kmem_cache_alloc+0xc0/0xce
[<c018bc1a>] locks_alloc_lock+0x12/0x14
[<c018c7ce>] __posix_lock_file_conf+0x64/0x439
[<c018cbb3>] posix_lock_file+0x10/0x12
[<c018d8c4>] fcntl_setlk+0x114/0x20e
[<c0189a94>] do_fcntl+0xf0/0x140
[<c0189bad>] sys_fcntl64+0x6e/0x82
[<c0105f3d>] sysenter_past_esp+0x56/0x8d
[<b7f10410>] 0xb7f10410
[<c0106f5a>] show_trace+0xd/0xf
[<c010702c>] dump_stack+0x17/0x19
[<c01fedbd>] __spin_lock_debug+0x86/0x93
[<c01fee28>] _raw_spin_lock+0x5e/0x73
[<c02fa771>] _spin_lock+0x2a/0x2f
[<c0176877>] memleak_alloc+0x17b/0x277
[<c01745a6>] kmem_cache_alloc+0xc0/0xce
[<c018bc1a>] locks_alloc_lock+0x12/0x14
[<c018c7ce>] __posix_lock_file_conf+0x64/0x439
[<c018cbb3>] posix_lock_file+0x10/0x12
[<c018d8c4>] fcntl_setlk+0x114/0x20e
[<c0189a94>] do_fcntl+0xf0/0x140
[<c0189bad>] sys_fcntl64+0x6e/0x82
[<c0105f3d>] sysenter_past_esp+0x56/0x8d

A few minutes later system has died. list *0xc0177468 in gdb shows nothing.

(gdb) l *0xc0177468
0xc0177468 is in memleak_scan (include2/asm/processor.h:1276).
Line number 1271 out of range; include2/asm/processor.h has 746 lines.

(gdb) l *memleak_scan+0x58/0x54e
0xc0177410 is in memleak_scan (/usr/src/linux-work8/mm/memleak.c:1252).
1247 spin_unlock_irqrestore(&memleak_lock, flags);
1248 }
1249
1250 /* Scan the memory and print the orphan pointers */
1251 static void memleak_scan(void)
1252 {
1253 unsigned long flags;
1254 struct memleak_pointer *pointer, *tmp;
1255 #ifdef CONFIG_DEBUG_MEMLEAK_TASK_STACKS
1256 struct task_struct *task;

http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/kml-dmesg3

>
> --
> Catalin
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-18 21:47:21

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> cat /sys/kernel/debug/memleak + LTP + patching repo = oops?
>
> BUG: unable to handle kernel paging request at virtual address 6b6b6b6b

I got this as well today on an 4-CPU ARM platform. I changed some list
traversal to the RCU variants but forgot to change the list_del/add. I
attach another patch which fixes this (haven't tested it extensively
yet).

Thanks.

--
Catalin


Attachments:
(No filename) (450.00 B)
new-locking.patch (27.14 kB)
Download all attachments

2006-08-18 22:49:53

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Catalin Marinas <[email protected]> wrote:
> On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > cat /sys/kernel/debug/memleak + LTP + patching repo = oops?
> >
> > BUG: unable to handle kernel paging request at virtual address 6b6b6b6b
>
> I got this as well today on an 4-CPU ARM platform. I changed some list
> traversal to the RCU variants but forgot to change the list_del/add. I
> attach another patch which fixes this (haven't tested it extensively
> yet).

Thanks.

I just added your "Fix memory leak in vc_resize/vc_allocate" patch to
my series file.

orphan pointer 0xc6110000 (size 12288):
c017480e: <__kmalloc>
c024dda4: <vc_resize>
c020ed9c: <fbcon_startup>
c0251028: <register_con_driver>
c02511e0: <take_over_console>
c020e21e: <fbcon_takeover>
c0212b08: <fbcon_fb_registered>
c0212ce1: <fbcon_event_notify>
orphan pointer 0xf55b0000 (size 8208):
c017480e: <__kmalloc>
c0211bb8: <fbcon_set_font>
c0251b17: <con_font_set>
c0251c7b: <con_font_op>
c0249a97: <vt_ioctl>
c024432e: <tty_ioctl>
c0189fd1: <do_ioctl>
c018a269: <vfs_ioctl>

This is interesting. Is it too large?
orphan pointer 0xfe09f000 (size 4194304):
c016cb9c: <__vmalloc_node>
c016cbb2: <__vmalloc>
c016cbc7: <vmalloc>
fdd3bbe1: <txInit>
fdd23b73: <init_jfs_fs>
c0149fea: <sys_init_module>

A simple way to reproduce:

while true
do
sudo mount -o loop -t jfs /home/fs-farm/jfs.img /mnt/fs-farm/jfs/
sudo umount /mnt/fs-farm/jfs/
done

A large collection of false positives :)
http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/ml_collection/ml1.tar

>
> Thanks.
>
> --
> Catalin
>
>
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-19 10:07:08

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> I just added your "Fix memory leak in vc_resize/vc_allocate" patch to
> my series file.
>
> orphan pointer 0xc6110000 (size 12288):
> c017480e: <__kmalloc>
> c024dda4: <vc_resize>
> c020ed9c: <fbcon_startup>
> c0251028: <register_con_driver>
> c02511e0: <take_over_console>
> c020e21e: <fbcon_takeover>
> c0212b08: <fbcon_fb_registered>
> c0212ce1: <fbcon_event_notify>
> orphan pointer 0xf55b0000 (size 8208):
> c017480e: <__kmalloc>
> c0211bb8: <fbcon_set_font>
> c0251b17: <con_font_set>
> c0251c7b: <con_font_op>
> c0249a97: <vt_ioctl>
> c024432e: <tty_ioctl>
> c0189fd1: <do_ioctl>
> c018a269: <vfs_ioctl>

The second one is probably a false positive as the stored pointer is
different from the one returned by kmalloc (there is some padding).
I'll add it to the kmemleak-false-positives patch.

The first one might be a real leak. Anotonino, any idea about this?

Thanks.

--
Catalin

2006-08-19 10:27:58

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> This is interesting. Is it too large?
> orphan pointer 0xfe09f000 (size 4194304):
> c016cb9c: <__vmalloc_node>
> c016cbb2: <__vmalloc>
> c016cbc7: <vmalloc>
> fdd3bbe1: <txInit>
> fdd23b73: <init_jfs_fs>
> c0149fea: <sys_init_module>

Looking at the code, I don't think it is too large. Anyway, it is just
a temporary false positive as it seems to only be present in some of
the files in your tarball.

> A large collection of false positives :)
> http://www.stardust.webpages.pl/files/o_bugs/kmemleak-0.9/ml_collection/ml1.tar

Thanks, but some of them look like real leaks :-) - delayacct_tsk_init
- I'll post a separate e-mail for this.

Have you seen any crashes or lockdep reports with the latest kmemleak patches?

--
Catalin

2006-08-19 10:32:14

by Michal Piotrowski

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 19/08/06, Catalin Marinas <[email protected]> wrote:
[snip]
> Have you seen any crashes or lockdep reports with the latest kmemleak patches?

12 hours of uptime - without any problems :).

>
> --
> Catalin
>

Regards,
Michal

--
Michal K. K. Piotrowski
LTG - Linux Testers Group
(http://www.stardust.webpages.pl/ltg/wiki/)

2006-08-19 11:21:56

by Antonino A. Daplas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On Sat, 2006-08-19 at 11:07 +0100, Catalin Marinas wrote:
> On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > I just added your "Fix memory leak in vc_resize/vc_allocate" patch to
> > my series file.
> >
> > orphan pointer 0xc6110000 (size 12288):
> > c017480e: <__kmalloc>
> > c024dda4: <vc_resize>
> > c020ed9c: <fbcon_startup>
> > c0251028: <register_con_driver>
> > c02511e0: <take_over_console>
> > c020e21e: <fbcon_takeover>
> > c0212b08: <fbcon_fb_registered>
> > c0212ce1: <fbcon_event_notify>
> > orphan pointer 0xf55b0000 (size 8208):
> > c017480e: <__kmalloc>
> > c0211bb8: <fbcon_set_font>
> > c0251b17: <con_font_set>
> > c0251c7b: <con_font_op>
> > c0249a97: <vt_ioctl>
> > c024432e: <tty_ioctl>
> > c0189fd1: <do_ioctl>
> > c018a269: <vfs_ioctl>
>
> The second one is probably a false positive as the stored pointer is
> different from the one returned by kmalloc (there is some padding).
> I'll add it to the kmemleak-false-positives patch.

Yes, I think it's triggered by the padding:

new_data += FONT_EXTRA_WORDS * sizeof(int);

The free'ing in fbcon_do_set_font() takes into account the padding:

if (old_data && (--REFCOUNT(old_data) == 0))
kfree(old_data - FONT_EXTRA_WORDS * sizeof(int));


> The first one might be a real leak. Anotonino, any idea about this?
>

On boot, vc->vc_screenbuf is allocated by alloc_bootmem() (vc->kmalloced
== 0), so yes, there's a leak in there . But I don't think we have a way
to deallocate this type of memory, so we just let it go.

Tony


2006-08-19 13:25:51

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 19/08/06, Antonino A. Daplas <[email protected]> wrote:
> On Sat, 2006-08-19 at 11:07 +0100, Catalin Marinas wrote:
> > On 18/08/06, Michal Piotrowski <[email protected]> wrote:
> > > orphan pointer 0xc6110000 (size 12288):
> > > c017480e: <__kmalloc>
> > > c024dda4: <vc_resize>
> > > c020ed9c: <fbcon_startup>
> > > c0251028: <register_con_driver>
> > > c02511e0: <take_over_console>
> > > c020e21e: <fbcon_takeover>
> > > c0212b08: <fbcon_fb_registered>
> > > c0212ce1: <fbcon_event_notify>
>
> On boot, vc->vc_screenbuf is allocated by alloc_bootmem() (vc->kmalloced
> == 0), so yes, there's a leak in there . But I don't think we have a way
> to deallocate this type of memory, so we just let it go.

It's not the leak caused by alloc_bootmem as kmemleak doesn't track
this kind of allocations. The block allocated via kmalloc in
vc_resize() is reported as unreferenced. However, I think that's a
false positive because it looks like the vc_screenbuf pointer is
stored in vc_cons[].d->vc_screenbuf which is allocated by
alloc_bootmem in con_init() and kmemleak doesn't track the
alloc_bootmem blocks. I'll fix kmemleak.

Thanks.

--
Catalin

2006-08-29 21:06:50

by Catalin Marinas

[permalink] [raw]
Subject: Re: [PATCH 2.6.18-rc4 00/10] Kernel memory leak detector 0.9

On 17/08/06, Catalin Marinas <[email protected]> wrote:
> it looks like it was caused by commit
> fc818301a8a39fedd7f0a71f878f29130c72193d where free_block() now calls
> slab_destroy() with l3->list_lock held.
[...]
> It leaves me with the options of either implementing my own memory
> allocator based on pages (including a simple hash table instead of
> radix tree) or fix the locking in kmemleak so that memory allocations
> happen without memleak_lock held. The latter is a bit complicated as
> well since any slab allocation causes a re-entrance into kmemleak.

FYI, I couldn't fix the locking dependency issues without modifying
the radix-tree implementation since even if you preload the tree
outside the memleak lock, it still tries to allocate memory in
radix_tree_insert.

I'll instead use a hash table (together with hlist) where I can
control the memory allocations and make more use of RCU. In my few
initial tests, the memory search with hash tables is about the same
speed (if not slightly faster) as the radix-tree implementation.

I will post a new kmemleak version by the end of this week.

--
Catalin