From 17666f02b118099028522dfc3df00a235700e216 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 30 Oct 2008 16:08:32 -0400 Subject: ftrace: nmi safe code modification Impact: fix crashes that can occur in NMI handlers, if their code is modified Modifying code is something that needs special care. On SMP boxes, if code that is being modified is also being executed on another CPU, that CPU will have undefined results. The dynamic ftrace uses kstop_machine to make the system act like a uniprocessor system. But this does not address NMIs, that can still run on other CPUs. One approach to handle this is to make all code that are used by NMIs not be traced. But NMIs can call notifiers that spread throughout the kernel and this will be very hard to maintain, and the chance of missing a function is very high. The approach that this patch takes is to have the NMIs modify the code if the modification is taking place. The way this works is that just writing to code executing on another CPU is not harmful if what is written is the same as what exists. Two buffers are used: an IP buffer and a "code" buffer. The steps that the patcher takes are: 1) Put in the instruction pointer into the IP buffer and the new code into the "code" buffer. 2) Set a flag that says we are modifying code 3) Wait for any running NMIs to finish. 4) Write the code 5) clear the flag. 6) Wait for any running NMIs to finish. If an NMI is executed, it will also write the pending code. Multiple writes are OK, because what is being written is the same. Then the patcher must wait for all running NMIs to finish before going to the next line that must be patched. This is basically the RCU approach to code modification. Thanks to Ingo Molnar for suggesting the idea, and to Arjan van de Ven for his guidence on what is safe and what is not. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/include/asm/ftrace.h | 15 +++++++++++++++ 1 file changed, 15 insertions(+) (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index 9e8bc29b8b1..f2ed6b704a7 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -17,6 +17,21 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr) */ return addr - 1; } + +#ifdef CONFIG_DYNAMIC_FTRACE +extern void ftrace_nmi_enter(void); +extern void ftrace_nmi_exit(void); +#else +#define ftrace_nmi_enter() do { } while (0) +#define ftrace_nmi_exit() do { } while (0) +#endif +#endif + +#else /* CONFIG_FUNCTION_TRACER */ + +#ifndef __ASSEMBLY__ +#define ftrace_nmi_enter() do { } while (0) +#define ftrace_nmi_exit() do { } while (0) #endif #endif /* CONFIG_FUNCTION_TRACER */ -- cgit v1.2.3-70-g09d2 From a26a2a27396c0a0877aa701f8f92d08ba550a6c9 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 31 Oct 2008 00:03:22 -0400 Subject: ftrace: nmi safe code clean ups Impact: cleanup This patch cleans up the NMI safe code for dynamic ftrace as suggested by Andrew Morton. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/arm/include/asm/ftrace.h | 4 ++-- arch/powerpc/include/asm/ftrace.h | 4 ++-- arch/sh/include/asm/ftrace.h | 4 ++-- arch/sparc/include/asm/ftrace.h | 4 ++-- arch/x86/include/asm/ftrace.h | 10 +++++----- arch/x86/kernel/ftrace.c | 16 ++++++++-------- include/linux/ftrace.h | 3 +++ kernel/trace/trace.c | 9 ++++----- 8 files changed, 28 insertions(+), 26 deletions(-) (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/arm/include/asm/ftrace.h b/arch/arm/include/asm/ftrace.h index d4c24a7a928..3f3a1d1508e 100644 --- a/arch/arm/include/asm/ftrace.h +++ b/arch/arm/include/asm/ftrace.h @@ -2,8 +2,8 @@ #define _ASM_ARM_FTRACE #ifndef __ASSEMBLY__ -#define ftrace_nmi_enter() do { } while (0) -#define ftrace_nmi_exit() do { } while (0) +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif #ifdef CONFIG_FUNCTION_TRACER diff --git a/arch/powerpc/include/asm/ftrace.h b/arch/powerpc/include/asm/ftrace.h index 7652755dc00..1cd72700fbc 100644 --- a/arch/powerpc/include/asm/ftrace.h +++ b/arch/powerpc/include/asm/ftrace.h @@ -2,8 +2,8 @@ #define _ASM_POWERPC_FTRACE #ifndef __ASSEMBLY__ -#define ftrace_nmi_enter() do { } while (0) -#define ftrace_nmi_exit() do { } while (0) +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif #ifdef CONFIG_FUNCTION_TRACER diff --git a/arch/sh/include/asm/ftrace.h b/arch/sh/include/asm/ftrace.h index cdf2cb0b9ff..31ada0370cb 100644 --- a/arch/sh/include/asm/ftrace.h +++ b/arch/sh/include/asm/ftrace.h @@ -2,8 +2,8 @@ #define __ASM_SH_FTRACE_H #ifndef __ASSEMBLY__ -#define ftrace_nmi_enter() do { } while (0) -#define ftrace_nmi_exit() do { } while (0) +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif #ifndef __ASSEMBLY__ diff --git a/arch/sparc/include/asm/ftrace.h b/arch/sparc/include/asm/ftrace.h index 33a95feeb13..62055ac0496 100644 --- a/arch/sparc/include/asm/ftrace.h +++ b/arch/sparc/include/asm/ftrace.h @@ -2,8 +2,8 @@ #define _ASM_SPARC64_FTRACE #ifndef __ASSEMBLY__ -#define ftrace_nmi_enter() do { } while (0) -#define ftrace_nmi_exit() do { } while (0) +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif #ifdef CONFIG_MCOUNT diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index f2ed6b704a7..a23468194b8 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -22,16 +22,16 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr) extern void ftrace_nmi_enter(void); extern void ftrace_nmi_exit(void); #else -#define ftrace_nmi_enter() do { } while (0) -#define ftrace_nmi_exit() do { } while (0) -#endif +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif +#endif /* __ASSEMBLY__ */ #else /* CONFIG_FUNCTION_TRACER */ #ifndef __ASSEMBLY__ -#define ftrace_nmi_enter() do { } while (0) -#define ftrace_nmi_exit() do { } while (0) +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif #endif /* CONFIG_FUNCTION_TRACER */ diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 6685b0fc1b4..69149337f2f 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -67,7 +67,7 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) * * Two buffers are added: An IP buffer and a "code" buffer. * - * 1) Put in the instruction pointer into the IP buffer + * 1) Put the instruction pointer into the IP buffer * and the new code into the "code" buffer. * 2) Set a flag that says we are modifying code * 3) Wait for any running NMIs to finish. @@ -85,14 +85,14 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) * are the same as what exists. */ -static atomic_t in_nmi; -static int mod_code_status; -static int mod_code_write; -static void *mod_code_ip; -static void *mod_code_newcode; +static atomic_t in_nmi = ATOMIC_INIT(0); +static int mod_code_status; /* holds return value of text write */ +static int mod_code_write; /* set when NMI should do the write */ +static void *mod_code_ip; /* holds the IP to write to */ +static void *mod_code_newcode; /* holds the text to write to the IP */ -static int nmi_wait_count; -static atomic_t nmi_update_count; +static unsigned nmi_wait_count; +static atomic_t nmi_update_count = ATOMIC_INIT(0); int ftrace_arch_read_dyn_info(char *buf, int size) { diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 703eb53cfa2..22240dfe912 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -74,6 +74,9 @@ extern void ftrace_caller(void); extern void ftrace_call(void); extern void mcount_call(void); +/* May be defined in arch */ +extern int ftrace_arch_read_dyn_info(char *buf, int size); + /** * ftrace_modify_code - modify code segment * @ip: the address of the code segment diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index bc36febc077..7f86067d760 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2815,10 +2815,6 @@ static struct file_operations tracing_mark_fops = { #ifdef CONFIG_DYNAMIC_FTRACE -#define DYN_INFO_BUF_SIZE 1023 -static char ftrace_dyn_info_buffer[DYN_INFO_BUF_SIZE+1]; -static DEFINE_MUTEX(dyn_info_mutex); - int __weak ftrace_arch_read_dyn_info(char *buf, int size) { return 0; @@ -2828,14 +2824,17 @@ static ssize_t tracing_read_dyn_info(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) { + static char ftrace_dyn_info_buffer[1024]; + static DEFINE_MUTEX(dyn_info_mutex); unsigned long *p = filp->private_data; char *buf = ftrace_dyn_info_buffer; + int size = ARRAY_SIZE(ftrace_dyn_info_buffer); int r; mutex_lock(&dyn_info_mutex); r = sprintf(buf, "%ld ", *p); - r += ftrace_arch_read_dyn_info(buf+r, DYN_INFO_BUF_SIZE-r); + r += ftrace_arch_read_dyn_info(buf+r, (size-1)-r); buf[r++] = '\n'; r = simple_read_from_buffer(ubuf, cnt, ppos, buf, r); -- cgit v1.2.3-70-g09d2 From 7e5e26a3d8ac4bcadb380073dc9604c07a9a6198 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 31 Oct 2008 09:36:38 -0400 Subject: ftrace: fix hardirq header for non ftrace archs Impact: build fix for non-ftrace architectures Not all archs implement ftrace, and therefore do not have an asm/ftrace.h. This patch corrects the problem. The ftrace_nmi_enter/exit now must be defined for all archs that implement dynamic ftrace. Currently, only x86 does. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/arm/include/asm/ftrace.h | 5 ----- arch/powerpc/include/asm/ftrace.h | 5 ----- arch/sh/include/asm/ftrace.h | 5 ----- arch/sparc/include/asm/ftrace.h | 5 ----- arch/x86/include/asm/ftrace.h | 16 ---------------- include/linux/ftrace.h | 5 ++++- include/linux/hardirq.h | 2 +- 7 files changed, 5 insertions(+), 38 deletions(-) (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/arm/include/asm/ftrace.h b/arch/arm/include/asm/ftrace.h index 3f3a1d1508e..39c8bc1a006 100644 --- a/arch/arm/include/asm/ftrace.h +++ b/arch/arm/include/asm/ftrace.h @@ -1,11 +1,6 @@ #ifndef _ASM_ARM_FTRACE #define _ASM_ARM_FTRACE -#ifndef __ASSEMBLY__ -static inline void ftrace_nmi_enter(void) { } -static inline void ftrace_nmi_exit(void) { } -#endif - #ifdef CONFIG_FUNCTION_TRACER #define MCOUNT_ADDR ((long)(mcount)) #define MCOUNT_INSN_SIZE 4 /* sizeof mcount call */ diff --git a/arch/powerpc/include/asm/ftrace.h b/arch/powerpc/include/asm/ftrace.h index 1cd72700fbc..b298f7a631e 100644 --- a/arch/powerpc/include/asm/ftrace.h +++ b/arch/powerpc/include/asm/ftrace.h @@ -1,11 +1,6 @@ #ifndef _ASM_POWERPC_FTRACE #define _ASM_POWERPC_FTRACE -#ifndef __ASSEMBLY__ -static inline void ftrace_nmi_enter(void) { } -static inline void ftrace_nmi_exit(void) { } -#endif - #ifdef CONFIG_FUNCTION_TRACER #define MCOUNT_ADDR ((long)(_mcount)) #define MCOUNT_INSN_SIZE 4 /* sizeof mcount call */ diff --git a/arch/sh/include/asm/ftrace.h b/arch/sh/include/asm/ftrace.h index 31ada0370cb..3aed362c946 100644 --- a/arch/sh/include/asm/ftrace.h +++ b/arch/sh/include/asm/ftrace.h @@ -1,11 +1,6 @@ #ifndef __ASM_SH_FTRACE_H #define __ASM_SH_FTRACE_H -#ifndef __ASSEMBLY__ -static inline void ftrace_nmi_enter(void) { } -static inline void ftrace_nmi_exit(void) { } -#endif - #ifndef __ASSEMBLY__ extern void mcount(void); #endif diff --git a/arch/sparc/include/asm/ftrace.h b/arch/sparc/include/asm/ftrace.h index 62055ac0496..d27716cd38c 100644 --- a/arch/sparc/include/asm/ftrace.h +++ b/arch/sparc/include/asm/ftrace.h @@ -1,11 +1,6 @@ #ifndef _ASM_SPARC64_FTRACE #define _ASM_SPARC64_FTRACE -#ifndef __ASSEMBLY__ -static inline void ftrace_nmi_enter(void) { } -static inline void ftrace_nmi_exit(void) { } -#endif - #ifdef CONFIG_MCOUNT #define MCOUNT_ADDR ((long)(_mcount)) #define MCOUNT_INSN_SIZE 4 /* sizeof mcount call */ diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index a23468194b8..f8173ed1c97 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -17,23 +17,7 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr) */ return addr - 1; } - -#ifdef CONFIG_DYNAMIC_FTRACE -extern void ftrace_nmi_enter(void); -extern void ftrace_nmi_exit(void); -#else -static inline void ftrace_nmi_enter(void) { } -static inline void ftrace_nmi_exit(void) { } -#endif #endif /* __ASSEMBLY__ */ - -#else /* CONFIG_FUNCTION_TRACER */ - -#ifndef __ASSEMBLY__ -static inline void ftrace_nmi_enter(void) { } -static inline void ftrace_nmi_exit(void) { } -#endif - #endif /* CONFIG_FUNCTION_TRACER */ #endif /* _ASM_X86_FTRACE_H */ diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index e46a7b34037..0ad1b48aea6 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -44,7 +44,6 @@ static inline void ftrace_kill(void) { } #endif /* CONFIG_FUNCTION_TRACER */ #ifdef CONFIG_DYNAMIC_FTRACE - enum { FTRACE_FL_FREE = (1 << 0), FTRACE_FL_FAILED = (1 << 1), @@ -105,6 +104,8 @@ extern void ftrace_release(void *start, unsigned long size); extern void ftrace_disable_daemon(void); extern void ftrace_enable_daemon(void); +extern void ftrace_nmi_enter(void); +extern void ftrace_nmi_exit(void); #else # define skip_trace(ip) ({ 0; }) @@ -113,6 +114,8 @@ extern void ftrace_enable_daemon(void); # define ftrace_disable_daemon() do { } while (0) # define ftrace_enable_daemon() do { } while (0) static inline void ftrace_release(void *start, unsigned long size) { } +static inline void ftrace_nmi_enter(void) { } +static inline void ftrace_nmi_exit(void) { } #endif /* CONFIG_DYNAMIC_FTRACE */ /* totally disable ftrace - can not re-enable after this */ diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h index 0087cb43bec..ffc16ab5a87 100644 --- a/include/linux/hardirq.h +++ b/include/linux/hardirq.h @@ -4,8 +4,8 @@ #include #include #include +#include #include -#include #include /* -- cgit v1.2.3-70-g09d2 From caf4b323b02a16c92fba449952ac6515ddc76d7a Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 11 Nov 2008 07:03:45 +0100 Subject: tracing, x86: add low level support for ftrace return tracing Impact: add infrastructure for function-return tracing Add low level support for ftrace return tracing. This plug-in stores return addresses on the thread_info structure of the current task. The index of the current return address is initialized when the task is the first one (init) and when a process forks (the child). It is not needed when a task does a sys_execve because after this syscall, it still needs to return on the kernel functions it called. Note that the code of return_to_handler has been suggested by Steven Rostedt as almost all of the ideas of improvements in this V3. For purpose of security, arch/x86/kernel/process_32.c is not traced because __switch_to() changes the current task during its execution. That could cause inconsistency in the stored return address of this function even if I didn't have any crash after testing with tracing on this function enabled. Signed-off-by: Frederic Weisbecker Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 1 + arch/x86/include/asm/ftrace.h | 26 ++++++ arch/x86/include/asm/thread_info.h | 24 +++++ arch/x86/kernel/Makefile | 6 ++ arch/x86/kernel/entry_32.S | 33 +++++++ arch/x86/kernel/ftrace.c | 181 +++++++++++++++++++++++++++++++++++-- include/linux/ftrace.h | 20 ++++ include/linux/ftrace_irq.h | 2 +- include/linux/sched.h | 11 +++ kernel/Makefile | 4 + 10 files changed, 300 insertions(+), 8 deletions(-) (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 27b8a3a3991..ca91e50bdb1 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -11,6 +11,7 @@ config 64BIT config X86_32 def_bool !64BIT + select HAVE_FUNCTION_RET_TRACER config X86_64 def_bool 64BIT diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index f8173ed1c97..9b6a1fa19e7 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -20,4 +20,30 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr) #endif /* __ASSEMBLY__ */ #endif /* CONFIG_FUNCTION_TRACER */ +#ifdef CONFIG_FUNCTION_RET_TRACER +#define FTRACE_RET_STACK_SIZE 20 + +#ifndef __ASSEMBLY__ + +/* + * Stack of return addresses for functions + * of a thread. + * Used in struct thread_info + */ +struct ftrace_ret_stack { + unsigned long ret; + unsigned long func; + unsigned long long calltime; +}; + +/* + * Primary handler of a function return. + * It relays on ftrace_return_to_handler. + * Defined in entry32.S + */ +extern void return_to_handler(void); + +#endif /* __ASSEMBLY__ */ +#endif /* CONFIG_FUNCTION_RET_TRACER */ + #endif /* _ASM_X86_FTRACE_H */ diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h index e44d379faad..a71158369fd 100644 --- a/arch/x86/include/asm/thread_info.h +++ b/arch/x86/include/asm/thread_info.h @@ -20,6 +20,7 @@ struct task_struct; struct exec_domain; #include +#include struct thread_info { struct task_struct *task; /* main task structure */ @@ -38,8 +39,30 @@ struct thread_info { */ __u8 supervisor_stack[0]; #endif + +#ifdef CONFIG_FUNCTION_RET_TRACER + /* Index of current stored adress in ret_stack */ + int curr_ret_stack; + /* Stack of return addresses for return function tracing */ + struct ftrace_ret_stack ret_stack[FTRACE_RET_STACK_SIZE]; +#endif }; +#ifdef CONFIG_FUNCTION_RET_TRACER +#define INIT_THREAD_INFO(tsk) \ +{ \ + .task = &tsk, \ + .exec_domain = &default_exec_domain, \ + .flags = 0, \ + .cpu = 0, \ + .preempt_count = 1, \ + .addr_limit = KERNEL_DS, \ + .restart_block = { \ + .fn = do_no_restart_syscall, \ + }, \ + .curr_ret_stack = -1,\ +} +#else #define INIT_THREAD_INFO(tsk) \ { \ .task = &tsk, \ @@ -52,6 +75,7 @@ struct thread_info { .fn = do_no_restart_syscall, \ }, \ } +#endif #define init_thread_info (init_thread_union.thread_info) #define init_stack (init_thread_union.stack) diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index e489ff9cb3e..1d8ed95da84 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -14,6 +14,11 @@ CFLAGS_REMOVE_paravirt-spinlocks.o = -pg CFLAGS_REMOVE_ftrace.o = -pg endif +ifdef CONFIG_FUNCTION_RET_TRACER +# Don't trace __switch_to() but let it for function tracer +CFLAGS_REMOVE_process_32.o = -pg +endif + # # vsyscalls (which work on the user stack) should have # no stack-protector checks: @@ -65,6 +70,7 @@ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o +obj-$(CONFIG_FUNCTION_RET_TRACER) += ftrace.o obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 9134de814c9..9a0ac85946d 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1188,6 +1188,10 @@ ENTRY(mcount) cmpl $ftrace_stub, ftrace_trace_function jnz trace +#ifdef CONFIG_FUNCTION_RET_TRACER + cmpl $ftrace_stub, ftrace_function_return + jnz trace_return +#endif .globl ftrace_stub ftrace_stub: ret @@ -1206,8 +1210,37 @@ trace: popl %edx popl %ecx popl %eax + jmp ftrace_stub +#ifdef CONFIG_FUNCTION_RET_TRACER +trace_return: + pushl %eax + pushl %ecx + pushl %edx + movl 0xc(%esp), %eax + pushl %eax + lea 0x4(%ebp), %eax + pushl %eax + call prepare_ftrace_return + addl $8, %esp + popl %edx + popl %ecx + popl %eax jmp ftrace_stub + +.globl return_to_handler +return_to_handler: + pushl $0 + pushl %eax + pushl %ecx + pushl %edx + call ftrace_return_to_handler + movl %eax, 0xc(%esp) + popl %edx + popl %ecx + popl %eax + ret +#endif /* CONFIG_FUNCTION_RET_TRACER */ END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* CONFIG_FUNCTION_TRACER */ diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 69149337f2f..d68033bba22 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -18,10 +18,173 @@ #include #include +#include #include +#include -static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; + +#ifdef CONFIG_FUNCTION_RET_TRACER + +/* + * These functions are picked from those used on + * this page for dynamic ftrace. They have been + * simplified to ignore all traces in NMI context. + */ +static atomic_t in_nmi; + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); +} + +void ftrace_nmi_exit(void) +{ + atomic_dec(&in_nmi); +} + +/* + * Synchronize accesses to return adresses stack with + * interrupts. + */ +static raw_spinlock_t ret_stack_lock; + +/* Add a function return address to the trace stack on thread info.*/ +static int push_return_trace(unsigned long ret, unsigned long long time, + unsigned long func) +{ + int index; + struct thread_info *ti; + unsigned long flags; + int err = 0; + + raw_local_irq_save(flags); + __raw_spin_lock(&ret_stack_lock); + + ti = current_thread_info(); + /* The return trace stack is full */ + if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) { + err = -EBUSY; + goto out; + } + + index = ++ti->curr_ret_stack; + ti->ret_stack[index].ret = ret; + ti->ret_stack[index].func = func; + ti->ret_stack[index].calltime = time; + +out: + __raw_spin_unlock(&ret_stack_lock); + raw_local_irq_restore(flags); + return err; +} + +/* Retrieve a function return address to the trace stack on thread info.*/ +static void pop_return_trace(unsigned long *ret, unsigned long long *time, + unsigned long *func) +{ + struct thread_info *ti; + int index; + unsigned long flags; + + raw_local_irq_save(flags); + __raw_spin_lock(&ret_stack_lock); + + ti = current_thread_info(); + index = ti->curr_ret_stack; + *ret = ti->ret_stack[index].ret; + *func = ti->ret_stack[index].func; + *time = ti->ret_stack[index].calltime; + ti->curr_ret_stack--; + + __raw_spin_unlock(&ret_stack_lock); + raw_local_irq_restore(flags); +} + +/* + * Send the trace to the ring-buffer. + * @return the original return address. + */ +unsigned long ftrace_return_to_handler(void) +{ + struct ftrace_retfunc trace; + pop_return_trace(&trace.ret, &trace.calltime, &trace.func); + trace.rettime = cpu_clock(raw_smp_processor_id()); + ftrace_function_return(&trace); + + return trace.ret; +} + +/* + * Hook the return address and push it in the stack of return addrs + * in current thread info. + */ +asmlinkage +void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) +{ + unsigned long old; + unsigned long long calltime; + int faulted; + unsigned long return_hooker = (unsigned long) + &return_to_handler; + + /* Nmi's are currently unsupported */ + if (atomic_read(&in_nmi)) + return; + + /* + * Protect against fault, even if it shouldn't + * happen. This tool is too much intrusive to + * ignore such a protection. + */ + asm volatile( + "1: movl (%[parent_old]), %[old]\n" + "2: movl %[return_hooker], (%[parent_replaced])\n" + " movl $0, %[faulted]\n" + + ".section .fixup, \"ax\"\n" + "3: movl $1, %[faulted]\n" + ".previous\n" + + ".section __ex_table, \"a\"\n" + " .long 1b, 3b\n" + " .long 2b, 3b\n" + ".previous\n" + + : [parent_replaced] "=rm" (parent), [old] "=r" (old), + [faulted] "=r" (faulted) + : [parent_old] "0" (parent), [return_hooker] "r" (return_hooker) + : "memory" + ); + + if (WARN_ON(faulted)) { + unregister_ftrace_return(); + return; + } + + if (WARN_ON(!__kernel_text_address(old))) { + unregister_ftrace_return(); + *parent = old; + return; + } + + calltime = cpu_clock(raw_smp_processor_id()); + + if (push_return_trace(old, calltime, self_addr) == -EBUSY) + *parent = old; +} + +static int __init init_ftrace_function_return(void) +{ + ret_stack_lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; + return 0; +} +device_initcall(init_ftrace_function_return); + + +#endif + +#ifdef CONFIG_DYNAMIC_FTRACE union ftrace_code_union { char code[MCOUNT_INSN_SIZE]; @@ -31,17 +194,11 @@ union ftrace_code_union { } __attribute__((packed)); }; - static int ftrace_calc_offset(long ip, long addr) { return (int)(addr - ip); } -unsigned char *ftrace_nop_replace(void) -{ - return ftrace_nop; -} - unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static union ftrace_code_union calc; @@ -183,6 +340,15 @@ do_ftrace_mod_code(unsigned long ip, void *new_code) } + + +static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; + +unsigned char *ftrace_nop_replace(void) +{ + return ftrace_nop; +} + int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) @@ -292,3 +458,4 @@ int __init ftrace_dyn_arch_init(void *data) return 0; } +#endif diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 1f5608c1102..dcbbf72a88b 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -267,6 +267,26 @@ ftrace_init_module(unsigned long *start, unsigned long *end) { } #endif +/* + * Structure that defines a return function trace. + */ +struct ftrace_retfunc { + unsigned long ret; /* Return address */ + unsigned long func; /* Current function */ + unsigned long long calltime; + unsigned long long rettime; +}; + +#ifdef CONFIG_FUNCTION_RET_TRACER +/* Type of a callback handler of tracing return function */ +typedef void (*trace_function_return_t)(struct ftrace_retfunc *); + +extern void register_ftrace_return(trace_function_return_t func); +/* The current handler in use */ +extern trace_function_return_t ftrace_function_return; +extern void unregister_ftrace_return(void); +#endif + /* * Structure which defines the trace of an initcall. * You don't have to fill the func field since it is diff --git a/include/linux/ftrace_irq.h b/include/linux/ftrace_irq.h index b1299d6729f..0b4df55d7a7 100644 --- a/include/linux/ftrace_irq.h +++ b/include/linux/ftrace_irq.h @@ -2,7 +2,7 @@ #define _LINUX_FTRACE_IRQ_H -#ifdef CONFIG_DYNAMIC_FTRACE +#if defined(CONFIG_DYNAMIC_FTRACE) || defined(CONFIG_FUNCTION_RET_TRACER) extern void ftrace_nmi_enter(void); extern void ftrace_nmi_exit(void); #else diff --git a/include/linux/sched.h b/include/linux/sched.h index 295b7c756ca..df77abe860c 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2005,6 +2005,17 @@ static inline void setup_thread_stack(struct task_struct *p, struct task_struct { *task_thread_info(p) = *task_thread_info(org); task_thread_info(p)->task = p; + +#ifdef CONFIG_FUNCTION_RET_TRACER + /* + * When fork() creates a child process, this function is called. + * But the child task may not inherit the return adresses traced + * by the return function tracer because it will directly execute + * in userspace and will not return to kernel functions its parent + * used. + */ + task_thread_info(p)->curr_ret_stack = -1; +#endif } static inline unsigned long *end_of_stack(struct task_struct *p) diff --git a/kernel/Makefile b/kernel/Makefile index 9a3ec66a9d8..af3be57acbb 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -23,6 +23,10 @@ CFLAGS_REMOVE_cgroup-debug.o = -pg CFLAGS_REMOVE_sched_clock.o = -pg CFLAGS_REMOVE_sched.o = -mno-spe -pg endif +ifdef CONFIG_FUNCTION_RET_TRACER +CFLAGS_REMOVE_extable.o = -pg # For __kernel_text_address() +CFLAGS_REMOVE_module.o = -pg # For __module_text_address() +endif obj-$(CONFIG_FREEZER) += freezer.o obj-$(CONFIG_PROFILING) += profile.o -- cgit v1.2.3-70-g09d2 From 31e889098a80ceb3e9e3c555d522b2686a6663c6 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 14 Nov 2008 16:21:19 -0800 Subject: ftrace: pass module struct to arch dynamic ftrace functions Impact: allow archs more flexibility on dynamic ftrace implementations Dynamic ftrace has largly been developed on x86. Since x86 does not have the same limitations as other architectures, the ftrace interaction between the generic code and the architecture specific code was not flexible enough to handle some of the issues that other architectures have. Most notably, module trampolines. Due to the limited branch distance that archs make in calling kernel core code from modules, the module load code must create a trampoline to jump to what will make the larger jump into core kernel code. The problem arises when this happens to a call to mcount. Ftrace checks all code before modifying it and makes sure the current code is what it expects. Right now, there is not enough information to handle modifying module trampolines. This patch changes the API between generic dynamic ftrace code and the arch dependent code. There is now two functions for modifying code: ftrace_make_nop(mod, rec, addr) - convert the code at rec->ip into a nop, where the original text is calling addr. (mod is the module struct if called by module init) ftrace_make_caller(rec, addr) - convert the code rec->ip that should be a nop into a caller to addr. The record "rec" now has a new field called "arch" where the architecture can add any special attributes to each call site record. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/include/asm/ftrace.h | 8 ++++++ arch/x86/kernel/ftrace.c | 29 +++++++++++++++++--- include/linux/ftrace.h | 53 +++++++++++++++++++++++++++--------- kernel/module.c | 2 +- kernel/trace/ftrace.c | 62 +++++++++++++++++-------------------------- 5 files changed, 100 insertions(+), 54 deletions(-) (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index 9b6a1fa19e7..2bb43b433e0 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -17,6 +17,14 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr) */ return addr - 1; } + +#ifdef CONFIG_DYNAMIC_FTRACE + +struct dyn_arch_ftrace { + /* No extra data needed for x86 */ +}; + +#endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* __ASSEMBLY__ */ #endif /* CONFIG_FUNCTION_TRACER */ diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index fe832738e1e..762222ad138 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -166,7 +166,7 @@ static int ftrace_calc_offset(long ip, long addr) return (int)(addr - ip); } -unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) +static unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) { static union ftrace_code_union calc; @@ -311,12 +311,12 @@ do_ftrace_mod_code(unsigned long ip, void *new_code) static unsigned char ftrace_nop[MCOUNT_INSN_SIZE]; -unsigned char *ftrace_nop_replace(void) +static unsigned char *ftrace_nop_replace(void) { return ftrace_nop; } -int +static int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) { @@ -349,6 +349,29 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return 0; } +int ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *new, *old; + unsigned long ip = rec->ip; + + old = ftrace_call_replace(ip, addr); + new = ftrace_nop_replace(); + + return ftrace_modify_code(rec->ip, old, new); +} + +int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned char *new, *old; + unsigned long ip = rec->ip; + + old = ftrace_nop_replace(); + new = ftrace_call_replace(ip, addr); + + return ftrace_modify_code(rec->ip, old, new); +} + int ftrace_update_ftrace_func(ftrace_func_t func) { unsigned long ip = (unsigned long)(&ftrace_call); diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 4fbc4a8b86a..166a2070ef6 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -74,6 +74,9 @@ static inline void ftrace_start(void) { } #endif /* CONFIG_FUNCTION_TRACER */ #ifdef CONFIG_DYNAMIC_FTRACE +/* asm/ftrace.h must be defined for archs supporting dynamic ftrace */ +#include + enum { FTRACE_FL_FREE = (1 << 0), FTRACE_FL_FAILED = (1 << 1), @@ -88,6 +91,7 @@ struct dyn_ftrace { struct list_head list; unsigned long ip; /* address of mcount call-site */ unsigned long flags; + struct dyn_arch_ftrace arch; }; int ftrace_force_update(void); @@ -95,22 +99,40 @@ void ftrace_set_filter(unsigned char *buf, int len, int reset); /* defined in arch */ extern int ftrace_ip_converted(unsigned long ip); -extern unsigned char *ftrace_nop_replace(void); -extern unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr); extern int ftrace_dyn_arch_init(void *data); extern int ftrace_update_ftrace_func(ftrace_func_t func); extern void ftrace_caller(void); extern void ftrace_call(void); extern void mcount_call(void); -/* May be defined in arch */ -extern int ftrace_arch_read_dyn_info(char *buf, int size); +/** + * ftrace_make_nop - convert code into top + * @mod: module structure if called by module load initialization + * @rec: the mcount call site record + * @addr: the address that the call site should be calling + * + * This is a very sensitive operation and great care needs + * to be taken by the arch. The operation should carefully + * read the location, check to see if what is read is indeed + * what we expect it to be, and then on success of the compare, + * it should write to the location. + * + * The code segment at @rec->ip should be a caller to @addr + * + * Return must be: + * 0 on success + * -EFAULT on error reading the location + * -EINVAL on a failed compare of the contents + * -EPERM on error writing to the location + * Any other value will be considered a failure. + */ +extern int ftrace_make_nop(struct module *mod, + struct dyn_ftrace *rec, unsigned long addr); /** - * ftrace_modify_code - modify code segment - * @ip: the address of the code segment - * @old_code: the contents of what is expected to be there - * @new_code: the code to patch in + * ftrace_make_call - convert a nop call site into a call to addr + * @rec: the mcount call site record + * @addr: the address that the call site should call * * This is a very sensitive operation and great care needs * to be taken by the arch. The operation should carefully @@ -118,6 +140,8 @@ extern int ftrace_arch_read_dyn_info(char *buf, int size); * what we expect it to be, and then on success of the compare, * it should write to the location. * + * The code segment at @rec->ip should be a nop + * * Return must be: * 0 on success * -EFAULT on error reading the location @@ -125,8 +149,11 @@ extern int ftrace_arch_read_dyn_info(char *buf, int size); * -EPERM on error writing to the location * Any other value will be considered a failure. */ -extern int ftrace_modify_code(unsigned long ip, unsigned char *old_code, - unsigned char *new_code); +extern int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr); + + +/* May be defined in arch */ +extern int ftrace_arch_read_dyn_info(char *buf, int size); extern int skip_trace(unsigned long ip); @@ -259,11 +286,13 @@ static inline void ftrace_dump(void) { } #ifdef CONFIG_FTRACE_MCOUNT_RECORD extern void ftrace_init(void); -extern void ftrace_init_module(unsigned long *start, unsigned long *end); +extern void ftrace_init_module(struct module *mod, + unsigned long *start, unsigned long *end); #else static inline void ftrace_init(void) { } static inline void -ftrace_init_module(unsigned long *start, unsigned long *end) { } +ftrace_init_module(struct module *mod, + unsigned long *start, unsigned long *end) { } #endif diff --git a/kernel/module.c b/kernel/module.c index 1f4cc00e0c2..69791274e89 100644 --- a/kernel/module.c +++ b/kernel/module.c @@ -2201,7 +2201,7 @@ static noinline struct module *load_module(void __user *umod, /* sechdrs[0].sh_size is always zero */ mseg = section_objs(hdr, sechdrs, secstrings, "__mcount_loc", sizeof(*mseg), &num_mcount); - ftrace_init_module(mseg, mseg + num_mcount); + ftrace_init_module(mod, mseg, mseg + num_mcount); err = module_finalize(hdr, sechdrs, mod); if (err < 0) diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 3940c71ac2a..e9a5fbfce08 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -358,9 +358,7 @@ static void print_ip_ins(const char *fmt, unsigned char *p) printk(KERN_CONT "%s%02x", i ? ":" : "", p[i]); } -static void ftrace_bug(int failed, unsigned long ip, - unsigned char *expected, - unsigned char *replace) +static void ftrace_bug(int failed, unsigned long ip) { switch (failed) { case -EFAULT: @@ -372,9 +370,7 @@ static void ftrace_bug(int failed, unsigned long ip, FTRACE_WARN_ON_ONCE(1); pr_info("ftrace failed to modify "); print_ip_sym(ip); - print_ip_ins(" expected: ", expected); print_ip_ins(" actual: ", (unsigned char *)ip); - print_ip_ins(" replace: ", replace); printk(KERN_CONT "\n"); break; case -EPERM: @@ -392,8 +388,7 @@ static void ftrace_bug(int failed, unsigned long ip, #define FTRACE_ADDR ((long)(ftrace_caller)) static int -__ftrace_replace_code(struct dyn_ftrace *rec, - unsigned char *old, unsigned char *new, int enable) +__ftrace_replace_code(struct dyn_ftrace *rec, int enable) { unsigned long ip, fl; @@ -435,12 +430,10 @@ __ftrace_replace_code(struct dyn_ftrace *rec, * otherwise enable it! */ if (fl & FTRACE_FL_ENABLED) { - /* swap new and old */ - new = old; - old = ftrace_call_replace(ip, FTRACE_ADDR); + enable = 0; rec->flags &= ~FTRACE_FL_ENABLED; } else { - new = ftrace_call_replace(ip, FTRACE_ADDR); + enable = 1; rec->flags |= FTRACE_FL_ENABLED; } } else { @@ -453,10 +446,7 @@ __ftrace_replace_code(struct dyn_ftrace *rec, fl = rec->flags & (FTRACE_FL_NOTRACE | FTRACE_FL_ENABLED); if (fl == FTRACE_FL_NOTRACE) return 0; - - new = ftrace_call_replace(ip, FTRACE_ADDR); - } else - old = ftrace_call_replace(ip, FTRACE_ADDR); + } if (enable) { if (rec->flags & FTRACE_FL_ENABLED) @@ -469,21 +459,18 @@ __ftrace_replace_code(struct dyn_ftrace *rec, } } - return ftrace_modify_code(ip, old, new); + if (enable) + return ftrace_make_call(rec, FTRACE_ADDR); + else + return ftrace_make_nop(NULL, rec, FTRACE_ADDR); } static void ftrace_replace_code(int enable) { int i, failed; - unsigned char *new = NULL, *old = NULL; struct dyn_ftrace *rec; struct ftrace_page *pg; - if (enable) - old = ftrace_nop_replace(); - else - new = ftrace_nop_replace(); - for (pg = ftrace_pages_start; pg; pg = pg->next) { for (i = 0; i < pg->index; i++) { rec = &pg->records[i]; @@ -504,34 +491,30 @@ static void ftrace_replace_code(int enable) unfreeze_record(rec); } - failed = __ftrace_replace_code(rec, old, new, enable); + failed = __ftrace_replace_code(rec, enable); if (failed && (rec->flags & FTRACE_FL_CONVERTED)) { rec->flags |= FTRACE_FL_FAILED; if ((system_state == SYSTEM_BOOTING) || !core_kernel_text(rec->ip)) { ftrace_free_rec(rec); } else - ftrace_bug(failed, rec->ip, old, new); + ftrace_bug(failed, rec->ip); } } } } static int -ftrace_code_disable(struct dyn_ftrace *rec) +ftrace_code_disable(struct module *mod, struct dyn_ftrace *rec) { unsigned long ip; - unsigned char *nop, *call; int ret; ip = rec->ip; - nop = ftrace_nop_replace(); - call = ftrace_call_replace(ip, mcount_addr); - - ret = ftrace_modify_code(ip, call, nop); + ret = ftrace_make_nop(mod, rec, mcount_addr); if (ret) { - ftrace_bug(ret, ip, call, nop); + ftrace_bug(ret, ip); rec->flags |= FTRACE_FL_FAILED; return 0; } @@ -650,7 +633,7 @@ static cycle_t ftrace_update_time; static unsigned long ftrace_update_cnt; unsigned long ftrace_update_tot_cnt; -static int ftrace_update_code(void) +static int ftrace_update_code(struct module *mod) { struct dyn_ftrace *p, *t; cycle_t start, stop; @@ -667,7 +650,7 @@ static int ftrace_update_code(void) list_del_init(&p->list); /* convert record (i.e, patch mcount-call with NOP) */ - if (ftrace_code_disable(p)) { + if (ftrace_code_disable(mod, p)) { p->flags |= FTRACE_FL_CONVERTED; ftrace_update_cnt++; } else @@ -1309,7 +1292,8 @@ static __init int ftrace_init_debugfs(void) fs_initcall(ftrace_init_debugfs); -static int ftrace_convert_nops(unsigned long *start, +static int ftrace_convert_nops(struct module *mod, + unsigned long *start, unsigned long *end) { unsigned long *p; @@ -1325,18 +1309,19 @@ static int ftrace_convert_nops(unsigned long *start, /* disable interrupts to prevent kstop machine */ local_irq_save(flags); - ftrace_update_code(); + ftrace_update_code(mod); local_irq_restore(flags); mutex_unlock(&ftrace_start_lock); return 0; } -void ftrace_init_module(unsigned long *start, unsigned long *end) +void ftrace_init_module(struct module *mod, + unsigned long *start, unsigned long *end) { if (ftrace_disabled || start == end) return; - ftrace_convert_nops(start, end); + ftrace_convert_nops(mod, start, end); } extern unsigned long __start_mcount_loc[]; @@ -1366,7 +1351,8 @@ void __init ftrace_init(void) last_ftrace_enabled = ftrace_enabled = 1; - ret = ftrace_convert_nops(__start_mcount_loc, + ret = ftrace_convert_nops(NULL, + __start_mcount_loc, __stop_mcount_loc); return; -- cgit v1.2.3-70-g09d2 From f201ae2356c74bcae130b2177b3dca903ea98071 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sun, 23 Nov 2008 06:22:56 +0100 Subject: tracing/function-return-tracer: store return stack into task_struct and allocate it dynamically Impact: use deeper function tracing depth safely Some tests showed that function return tracing needed a more deeper depth of function calls. But it could be unsafe to store these return addresses to the stack. So these arrays will now be allocated dynamically into task_struct of current only when the tracer is activated. Typical scheme when tracer is activated: - allocate a return stack for each task in global list. - fork: allocate the return stack for the newly created task - exit: free return stack of current - idle init: same as fork I chose a default depth of 50. I don't have overruns anymore. Signed-off-by: Frederic Weisbecker Signed-off-by: Ingo Molnar --- arch/x86/include/asm/ftrace.h | 1 - arch/x86/include/asm/thread_info.h | 29 ------------ arch/x86/kernel/ftrace.c | 29 ++++++------ include/linux/ftrace.h | 5 ++ include/linux/sched.h | 23 +++++---- kernel/exit.c | 5 +- kernel/fork.c | 4 ++ kernel/sched.c | 3 ++ kernel/trace/ftrace.c | 96 +++++++++++++++++++++++++++++++++++++- 9 files changed, 137 insertions(+), 58 deletions(-) (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index 2bb43b433e0..754a3e082f9 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -29,7 +29,6 @@ struct dyn_arch_ftrace { #endif /* CONFIG_FUNCTION_TRACER */ #ifdef CONFIG_FUNCTION_RET_TRACER -#define FTRACE_RET_STACK_SIZE 20 #ifndef __ASSEMBLY__ diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h index e90e81ef6ab..0921b4018c1 100644 --- a/arch/x86/include/asm/thread_info.h +++ b/arch/x86/include/asm/thread_info.h @@ -40,36 +40,8 @@ struct thread_info { */ __u8 supervisor_stack[0]; #endif - -#ifdef CONFIG_FUNCTION_RET_TRACER - /* Index of current stored adress in ret_stack */ - int curr_ret_stack; - /* Stack of return addresses for return function tracing */ - struct ftrace_ret_stack ret_stack[FTRACE_RET_STACK_SIZE]; - /* - * Number of functions that haven't been traced - * because of depth overrun. - */ - atomic_t trace_overrun; -#endif }; -#ifdef CONFIG_FUNCTION_RET_TRACER -#define INIT_THREAD_INFO(tsk) \ -{ \ - .task = &tsk, \ - .exec_domain = &default_exec_domain, \ - .flags = 0, \ - .cpu = 0, \ - .preempt_count = 1, \ - .addr_limit = KERNEL_DS, \ - .restart_block = { \ - .fn = do_no_restart_syscall, \ - }, \ - .curr_ret_stack = -1,\ - .trace_overrun = ATOMIC_INIT(0) \ -} -#else #define INIT_THREAD_INFO(tsk) \ { \ .task = &tsk, \ @@ -82,7 +54,6 @@ struct thread_info { .fn = do_no_restart_syscall, \ }, \ } -#endif #define init_thread_info (init_thread_union.thread_info) #define init_stack (init_thread_union.stack) diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 356bb1eb6e9..bb137f7297e 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -350,19 +350,21 @@ static int push_return_trace(unsigned long ret, unsigned long long time, unsigned long func) { int index; - struct thread_info *ti = current_thread_info(); + + if (!current->ret_stack) + return -EBUSY; /* The return trace stack is full */ - if (ti->curr_ret_stack == FTRACE_RET_STACK_SIZE - 1) { - atomic_inc(&ti->trace_overrun); + if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) { + atomic_inc(¤t->trace_overrun); return -EBUSY; } - index = ++ti->curr_ret_stack; + index = ++current->curr_ret_stack; barrier(); - ti->ret_stack[index].ret = ret; - ti->ret_stack[index].func = func; - ti->ret_stack[index].calltime = time; + current->ret_stack[index].ret = ret; + current->ret_stack[index].func = func; + current->ret_stack[index].calltime = time; return 0; } @@ -373,13 +375,12 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, { int index; - struct thread_info *ti = current_thread_info(); - index = ti->curr_ret_stack; - *ret = ti->ret_stack[index].ret; - *func = ti->ret_stack[index].func; - *time = ti->ret_stack[index].calltime; - *overrun = atomic_read(&ti->trace_overrun); - ti->curr_ret_stack--; + index = current->curr_ret_stack; + *ret = current->ret_stack[index].ret; + *func = current->ret_stack[index].func; + *time = current->ret_stack[index].calltime; + *overrun = atomic_read(¤t->trace_overrun); + current->curr_ret_stack--; } /* diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index f7ba4ea5e12..2ba259b2def 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -323,6 +323,8 @@ struct ftrace_retfunc { }; #ifdef CONFIG_FUNCTION_RET_TRACER +#define FTRACE_RETFUNC_DEPTH 50 +#define FTRACE_RETSTACK_ALLOC_SIZE 32 /* Type of a callback handler of tracing return function */ typedef void (*trace_function_return_t)(struct ftrace_retfunc *); @@ -330,6 +332,9 @@ extern int register_ftrace_return(trace_function_return_t func); /* The current handler in use */ extern trace_function_return_t ftrace_function_return; extern void unregister_ftrace_return(void); + +extern void ftrace_retfunc_init_task(struct task_struct *t); +extern void ftrace_retfunc_exit_task(struct task_struct *t); #endif #endif /* _LINUX_FTRACE_H */ diff --git a/include/linux/sched.h b/include/linux/sched.h index c8e0db46420..bee1e93c95a 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1352,6 +1352,17 @@ struct task_struct { unsigned long default_timer_slack_ns; struct list_head *scm_work_list; +#ifdef CONFIG_FUNCTION_RET_TRACER + /* Index of current stored adress in ret_stack */ + int curr_ret_stack; + /* Stack of return addresses for return function tracing */ + struct ftrace_ret_stack *ret_stack; + /* + * Number of functions that haven't been traced + * because of depth overrun. + */ + atomic_t trace_overrun; +#endif }; /* @@ -2006,18 +2017,6 @@ static inline void setup_thread_stack(struct task_struct *p, struct task_struct { *task_thread_info(p) = *task_thread_info(org); task_thread_info(p)->task = p; - -#ifdef CONFIG_FUNCTION_RET_TRACER - /* - * When fork() creates a child process, this function is called. - * But the child task may not inherit the return adresses traced - * by the return function tracer because it will directly execute - * in userspace and will not return to kernel functions its parent - * used. - */ - task_thread_info(p)->curr_ret_stack = -1; - atomic_set(&task_thread_info(p)->trace_overrun, 0); -#endif } static inline unsigned long *end_of_stack(struct task_struct *p) diff --git a/kernel/exit.c b/kernel/exit.c index 35c8ec2ba03..b9d446329da 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -47,6 +47,7 @@ #include #include #include +#include #include #include @@ -1127,7 +1128,9 @@ NORET_TYPE void do_exit(long code) preempt_disable(); /* causes final put_task_struct in finish_task_switch(). */ tsk->state = TASK_DEAD; - +#ifdef CONFIG_FUNCTION_RET_TRACER + ftrace_retfunc_exit_task(tsk); +#endif schedule(); BUG(); /* Avoid "noreturn function does return". */ diff --git a/kernel/fork.c b/kernel/fork.c index ac62f43ee43..d1eb30e69cc 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -47,6 +47,7 @@ #include #include #include +#include #include #include #include @@ -1269,6 +1270,9 @@ static struct task_struct *copy_process(unsigned long clone_flags, total_forks++; spin_unlock(¤t->sighand->siglock); write_unlock_irq(&tasklist_lock); +#ifdef CONFIG_FUNCTION_RET_TRACER + ftrace_retfunc_init_task(p); +#endif proc_fork_connector(p); cgroup_post_fork(p); return p; diff --git a/kernel/sched.c b/kernel/sched.c index 4de56108c86..fb17205950d 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5901,6 +5901,9 @@ void __cpuinit init_idle(struct task_struct *idle, int cpu) * The idle tasks have their own, simple scheduling class: */ idle->sched_class = &idle_sched_class; +#ifdef CONFIG_FUNCTION_RET_TRACER + ftrace_retfunc_init_task(idle); +#endif } /* diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index f212da48668..90d99fb02ae 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1498,10 +1498,77 @@ ftrace_enable_sysctl(struct ctl_table *table, int write, #ifdef CONFIG_FUNCTION_RET_TRACER +static atomic_t ftrace_retfunc_active; + /* The callback that hooks the return of a function */ trace_function_return_t ftrace_function_return = (trace_function_return_t)ftrace_stub; + +/* Try to assign a return stack array on FTRACE_RETSTACK_ALLOC_SIZE tasks. */ +static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list) +{ + int i; + int ret = 0; + unsigned long flags; + int start = 0, end = FTRACE_RETSTACK_ALLOC_SIZE; + struct task_struct *g, *t; + + for (i = 0; i < FTRACE_RETSTACK_ALLOC_SIZE; i++) { + ret_stack_list[i] = kmalloc(FTRACE_RETFUNC_DEPTH + * sizeof(struct ftrace_ret_stack), + GFP_KERNEL); + if (!ret_stack_list[i]) { + start = 0; + end = i; + ret = -ENOMEM; + goto free; + } + } + + read_lock_irqsave(&tasklist_lock, flags); + do_each_thread(g, t) { + if (start == end) { + ret = -EAGAIN; + goto unlock; + } + + if (t->ret_stack == NULL) { + t->ret_stack = ret_stack_list[start++]; + t->curr_ret_stack = -1; + atomic_set(&t->trace_overrun, 0); + } + } while_each_thread(g, t); + +unlock: + read_unlock_irqrestore(&tasklist_lock, flags); +free: + for (i = start; i < end; i++) + kfree(ret_stack_list[i]); + return ret; +} + +/* Allocate a return stack for each task */ +static int start_return_tracing(void) +{ + struct ftrace_ret_stack **ret_stack_list; + int ret; + + ret_stack_list = kmalloc(FTRACE_RETSTACK_ALLOC_SIZE * + sizeof(struct ftrace_ret_stack *), + GFP_KERNEL); + + if (!ret_stack_list) + return -ENOMEM; + + do { + ret = alloc_retstack_tasklist(ret_stack_list); + } while (ret == -EAGAIN); + + kfree(ret_stack_list); + return ret; +} + int register_ftrace_return(trace_function_return_t func) { int ret = 0; @@ -1516,7 +1583,12 @@ int register_ftrace_return(trace_function_return_t func) ret = -EBUSY; goto out; } - + atomic_inc(&ftrace_retfunc_active); + ret = start_return_tracing(); + if (ret) { + atomic_dec(&ftrace_retfunc_active); + goto out; + } ftrace_tracing_type = FTRACE_TYPE_RETURN; ftrace_function_return = func; ftrace_startup(); @@ -1530,6 +1602,7 @@ void unregister_ftrace_return(void) { mutex_lock(&ftrace_sysctl_lock); + atomic_dec(&ftrace_retfunc_active); ftrace_function_return = (trace_function_return_t)ftrace_stub; ftrace_shutdown(); /* Restore normal tracing type */ @@ -1537,6 +1610,27 @@ void unregister_ftrace_return(void) mutex_unlock(&ftrace_sysctl_lock); } + +/* Allocate a return stack for newly created task */ +void ftrace_retfunc_init_task(struct task_struct *t) +{ + if (atomic_read(&ftrace_retfunc_active)) { + t->ret_stack = kmalloc(FTRACE_RETFUNC_DEPTH + * sizeof(struct ftrace_ret_stack), + GFP_KERNEL); + if (!t->ret_stack) + return; + t->curr_ret_stack = -1; + atomic_set(&t->trace_overrun, 0); + } else + t->ret_stack = NULL; +} + +void ftrace_retfunc_exit_task(struct task_struct *t) +{ + kfree(t->ret_stack); + t->ret_stack = NULL; +} #endif -- cgit v1.2.3-70-g09d2 From fb52607afcd0629776f1dc9e657647ceae81dd50 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Tue, 25 Nov 2008 21:07:04 +0100 Subject: tracing/function-return-tracer: change the name into function-graph-tracer Impact: cleanup This patch changes the name of the "return function tracer" into function-graph-tracer which is a more suitable name for a tracing which makes one able to retrieve the ordered call stack during the code flow. Signed-off-by: Frederic Weisbecker Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 2 +- arch/x86/include/asm/ftrace.h | 4 +- arch/x86/kernel/Makefile | 4 +- arch/x86/kernel/entry_32.S | 12 ++--- arch/x86/kernel/ftrace.c | 12 ++--- include/linux/ftrace.h | 24 ++++----- include/linux/ftrace_irq.h | 2 +- include/linux/sched.h | 2 +- kernel/Makefile | 2 +- kernel/fork.c | 4 +- kernel/sched.c | 2 +- kernel/trace/Kconfig | 19 ++++--- kernel/trace/Makefile | 2 +- kernel/trace/ftrace.c | 26 +++++----- kernel/trace/trace.c | 18 +++---- kernel/trace/trace.h | 12 ++--- kernel/trace/trace_functions_graph.c | 98 ++++++++++++++++++++++++++++++++++++ 17 files changed, 173 insertions(+), 72 deletions(-) create mode 100644 kernel/trace/trace_functions_graph.c (limited to 'arch/x86/include/asm/ftrace.h') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index e49a4fd718f..0842b112768 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -29,7 +29,7 @@ config X86 select HAVE_FTRACE_MCOUNT_RECORD select HAVE_DYNAMIC_FTRACE select HAVE_FUNCTION_TRACER - select HAVE_FUNCTION_RET_TRACER if X86_32 + select HAVE_FUNCTION_GRAPH_TRACER if X86_32 select HAVE_FUNCTION_TRACE_MCOUNT_TEST select HAVE_KVM if ((X86_32 && !X86_VOYAGER && !X86_VISWS && !X86_NUMAQ) || X86_64) select HAVE_ARCH_KGDB if !X86_VOYAGER diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h index 754a3e082f9..7e61b4ceb9a 100644 --- a/arch/x86/include/asm/ftrace.h +++ b/arch/x86/include/asm/ftrace.h @@ -28,7 +28,7 @@ struct dyn_arch_ftrace { #endif /* __ASSEMBLY__ */ #endif /* CONFIG_FUNCTION_TRACER */ -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER #ifndef __ASSEMBLY__ @@ -51,6 +51,6 @@ struct ftrace_ret_stack { extern void return_to_handler(void); #endif /* __ASSEMBLY__ */ -#endif /* CONFIG_FUNCTION_RET_TRACER */ +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ #endif /* _ASM_X86_FTRACE_H */ diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index af2bc36ca1c..64939a0c398 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -14,7 +14,7 @@ CFLAGS_REMOVE_paravirt-spinlocks.o = -pg CFLAGS_REMOVE_ftrace.o = -pg endif -ifdef CONFIG_FUNCTION_RET_TRACER +ifdef CONFIG_FUNCTION_GRAPH_TRACER # Don't trace __switch_to() but let it for function tracer CFLAGS_REMOVE_process_32.o = -pg endif @@ -70,7 +70,7 @@ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o nmi.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o obj-$(CONFIG_X86_REBOOTFIXUPS) += reboot_fixups_32.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o -obj-$(CONFIG_FUNCTION_RET_TRACER) += ftrace.o +obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o obj-$(CONFIG_KEXEC) += machine_kexec_$(BITS).o obj-$(CONFIG_KEXEC) += relocate_kernel_$(BITS).o crash.o obj-$(CONFIG_CRASH_DUMP) += crash_dump_$(BITS).o diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 74defe21ba4..2b1f0f081a6 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -1188,9 +1188,9 @@ ENTRY(mcount) cmpl $ftrace_stub, ftrace_trace_function jnz trace -#ifdef CONFIG_FUNCTION_RET_TRACER - cmpl $ftrace_stub, ftrace_function_return - jnz ftrace_return_caller +#ifdef CONFIG_FUNCTION_GRAPH_TRACER + cmpl $ftrace_stub, ftrace_graph_function + jnz ftrace_graph_caller #endif .globl ftrace_stub ftrace_stub: @@ -1215,8 +1215,8 @@ END(mcount) #endif /* CONFIG_DYNAMIC_FTRACE */ #endif /* CONFIG_FUNCTION_TRACER */ -#ifdef CONFIG_FUNCTION_RET_TRACER -ENTRY(ftrace_return_caller) +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +ENTRY(ftrace_graph_caller) cmpl $0, function_trace_stop jne ftrace_stub @@ -1230,7 +1230,7 @@ ENTRY(ftrace_return_caller) popl %ecx popl %eax ret -END(ftrace_return_caller) +END(ftrace_graph_caller) .globl return_to_handler return_to_handler: diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index bb137f7297e..3595a4c14ab 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -323,7 +323,7 @@ int __init ftrace_dyn_arch_init(void *data) } #endif -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER #ifndef CONFIG_DYNAMIC_FTRACE @@ -389,11 +389,11 @@ static void pop_return_trace(unsigned long *ret, unsigned long long *time, */ unsigned long ftrace_return_to_handler(void) { - struct ftrace_retfunc trace; + struct ftrace_graph_ret trace; pop_return_trace(&trace.ret, &trace.calltime, &trace.func, &trace.overrun); trace.rettime = cpu_clock(raw_smp_processor_id()); - ftrace_function_return(&trace); + ftrace_graph_function(&trace); return trace.ret; } @@ -440,12 +440,12 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) ); if (WARN_ON(faulted)) { - unregister_ftrace_return(); + unregister_ftrace_graph(); return; } if (WARN_ON(!__kernel_text_address(old))) { - unregister_ftrace_return(); + unregister_ftrace_graph(); *parent = old; return; } @@ -456,4 +456,4 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr) *parent = old; } -#endif /* CONFIG_FUNCTION_RET_TRACER */ +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h index 7854d87b97b..b4ac734ad8d 100644 --- a/include/linux/ftrace.h +++ b/include/linux/ftrace.h @@ -115,8 +115,8 @@ extern int ftrace_update_ftrace_func(ftrace_func_t func); extern void ftrace_caller(void); extern void ftrace_call(void); extern void mcount_call(void); -#ifdef CONFIG_FUNCTION_RET_TRACER -extern void ftrace_return_caller(void); +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +extern void ftrace_graph_caller(void); #endif /** @@ -315,7 +315,7 @@ ftrace_init_module(struct module *mod, /* * Structure that defines a return function trace. */ -struct ftrace_retfunc { +struct ftrace_graph_ret { unsigned long ret; /* Return address */ unsigned long func; /* Current function */ unsigned long long calltime; @@ -324,22 +324,22 @@ struct ftrace_retfunc { unsigned long overrun; }; -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER #define FTRACE_RETFUNC_DEPTH 50 #define FTRACE_RETSTACK_ALLOC_SIZE 32 /* Type of a callback handler of tracing return function */ -typedef void (*trace_function_return_t)(struct ftrace_retfunc *); +typedef void (*trace_function_graph_t)(struct ftrace_graph_ret *); -extern int register_ftrace_return(trace_function_return_t func); +extern int register_ftrace_graph(trace_function_graph_t func); /* The current handler in use */ -extern trace_function_return_t ftrace_function_return; -extern void unregister_ftrace_return(void); +extern trace_function_graph_t ftrace_graph_function; +extern void unregister_ftrace_graph(void); -extern void ftrace_retfunc_init_task(struct task_struct *t); -extern void ftrace_retfunc_exit_task(struct task_struct *t); +extern void ftrace_graph_init_task(struct task_struct *t); +extern void ftrace_graph_exit_task(struct task_struct *t); #else -static inline void ftrace_retfunc_init_task(struct task_struct *t) { } -static inline void ftrace_retfunc_exit_task(struct task_struct *t) { } +static inline void ftrace_graph_init_task(struct task_struct *t) { } +static inline void ftrace_graph_exit_task(struct task_struct *t) { } #endif #endif /* _LINUX_FTRACE_H */ diff --git a/include/linux/ftrace_irq.h b/include/linux/ftrace_irq.h index 0b4df55d7a7..366a054d0b0 100644 --- a/include/linux/ftrace_irq.h +++ b/include/linux/ftrace_irq.h @@ -2,7 +2,7 @@ #define _LINUX_FTRACE_IRQ_H -#if defined(CONFIG_DYNAMIC_FTRACE) || defined(CONFIG_FUNCTION_RET_TRACER) +#if defined(CONFIG_DYNAMIC_FTRACE) || defined(CONFIG_FUNCTION_GRAPH_TRACER) extern void ftrace_nmi_enter(void); extern void ftrace_nmi_exit(void); #else diff --git a/include/linux/sched.h b/include/linux/sched.h index d02a0ca70ee..7ad48f2a275 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1365,7 +1365,7 @@ struct task_struct { unsigned long default_timer_slack_ns; struct list_head *scm_work_list; -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER /* Index of current stored adress in ret_stack */ int curr_ret_stack; /* Stack of return addresses for return function tracing */ diff --git a/kernel/Makefile b/kernel/Makefile index 03a45e7e87b..703cf3b7389 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -21,7 +21,7 @@ CFLAGS_REMOVE_cgroup-debug.o = -pg CFLAGS_REMOVE_sched_clock.o = -pg CFLAGS_REMOVE_sched.o = -pg endif -ifdef CONFIG_FUNCTION_RET_TRACER +ifdef CONFIG_FUNCTION_GRAPH_TRACER CFLAGS_REMOVE_extable.o = -pg # For __kernel_text_address() CFLAGS_REMOVE_module.o = -pg # For __module_text_address() endif diff --git a/kernel/fork.c b/kernel/fork.c index d6e1a3205f6..5f82a999c03 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -140,7 +140,7 @@ void free_task(struct task_struct *tsk) prop_local_destroy_single(&tsk->dirties); free_thread_info(tsk->stack); rt_mutex_debug_task_free(tsk); - ftrace_retfunc_exit_task(tsk); + ftrace_graph_exit_task(tsk); free_task_struct(tsk); } EXPORT_SYMBOL(free_task); @@ -1271,7 +1271,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, total_forks++; spin_unlock(¤t->sighand->siglock); write_unlock_irq(&tasklist_lock); - ftrace_retfunc_init_task(p); + ftrace_graph_init_task(p); proc_fork_connector(p); cgroup_post_fork(p); return p; diff --git a/kernel/sched.c b/kernel/sched.c index 388d9db044a..52490bf6b88 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5901,7 +5901,7 @@ void __cpuinit init_idle(struct task_struct *idle, int cpu) * The idle tasks have their own, simple scheduling class: */ idle->sched_class = &idle_sched_class; - ftrace_retfunc_init_task(idle); + ftrace_graph_init_task(idle); } /* diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 620feadff67..eb9b901e077 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -12,7 +12,7 @@ config NOP_TRACER config HAVE_FUNCTION_TRACER bool -config HAVE_FUNCTION_RET_TRACER +config HAVE_FUNCTION_GRAPH_TRACER bool config HAVE_FUNCTION_TRACE_MCOUNT_TEST @@ -63,15 +63,18 @@ config FUNCTION_TRACER (the bootup default), then the overhead of the instructions is very small and not measurable even in micro-benchmarks. -config FUNCTION_RET_TRACER - bool "Kernel Function return Tracer" - depends on HAVE_FUNCTION_RET_TRACER +config FUNCTION_GRAPH_TRACER + bool "Kernel Function Graph Tracer" + depends on HAVE_FUNCTION_GRAPH_TRACER depends on FUNCTION_TRACER help - Enable the kernel to trace a function at its return. - It's first purpose is to trace the duration of functions. - This is done by setting the current return address on the thread - info structure of the current task. + Enable the kernel to trace a function at both its return + and its entry. + It's first purpose is to trace the duration of functions and + draw a call graph for each thread with some informations like + the return value. + This is done by setting the current return address on the current + task structure into a stack of calls. config IRQSOFF_TRACER bool "Interrupts-off Latency Tracer" diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index cef4bcb4e82..08c5fe6ddc0 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -29,7 +29,7 @@ obj-$(CONFIG_NOP_TRACER) += trace_nop.o obj-$(CONFIG_STACK_TRACER) += trace_stack.o obj-$(CONFIG_MMIOTRACE) += trace_mmiotrace.o obj-$(CONFIG_BOOT_TRACER) += trace_boot.o -obj-$(CONFIG_FUNCTION_RET_TRACER) += trace_functions_return.o +obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += trace_functions_graph.o obj-$(CONFIG_TRACE_BRANCH_PROFILING) += trace_branch.o obj-$(CONFIG_BTS_TRACER) += trace_bts.o diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 53042f118f2..9e19976af72 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -395,11 +395,11 @@ __ftrace_replace_code(struct dyn_ftrace *rec, int enable) unsigned long ip, fl; unsigned long ftrace_addr; -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER if (ftrace_tracing_type == FTRACE_TYPE_ENTER) ftrace_addr = (unsigned long)ftrace_caller; else - ftrace_addr = (unsigned long)ftrace_return_caller; + ftrace_addr = (unsigned long)ftrace_graph_caller; #else ftrace_addr = (unsigned long)ftrace_caller; #endif @@ -1496,13 +1496,13 @@ ftrace_enable_sysctl(struct ctl_table *table, int write, return ret; } -#ifdef CONFIG_FUNCTION_RET_TRACER +#ifdef CONFIG_FUNCTION_GRAPH_TRACER static atomic_t ftrace_retfunc_active; /* The callback that hooks the return of a function */ -trace_function_return_t ftrace_function_return = - (trace_function_return_t)ftrace_stub; +trace_function_graph_t ftrace_graph_function = + (trace_function_graph_t)ftrace_stub; /* Try to assign a return stack array on FTRACE_RETSTACK_ALLOC_SIZE tasks. */ @@ -1549,7 +1549,7 @@ free: } /* Allocate a return stack for each task */ -static int start_return_tracing(void) +static int start_graph_tracing(void) { struct ftrace_ret_stack **ret_stack_list; int ret; @@ -1569,7 +1569,7 @@ static int start_return_tracing(void) return ret; } -int register_ftrace_return(trace_function_return_t func) +int register_ftrace_graph(trace_function_graph_t func) { int ret = 0; @@ -1584,13 +1584,13 @@ int register_ftrace_return(trace_function_return_t func) goto out; } atomic_inc(&ftrace_retfunc_active); - ret = start_return_tracing(); + ret = start_graph_tracing(); if (ret) { atomic_dec(&ftrace_retfunc_active); goto out; } ftrace_tracing_type = FTRACE_TYPE_RETURN; - ftrace_function_return = func; + ftrace_graph_function = func; ftrace_startup(); out: @@ -1598,12 +1598,12 @@ out: return ret; } -void unregister_ftrace_return(void) +void unregister_ftrace_graph(void) { mutex_lock(&ftrace_sysctl_lock); atomic_dec(&ftrace_retfunc_active); - ftrace_function_return = (trace_function_return_t)ftrace_stub; + ftrace_graph_function = (trace_function_graph_t)ftrace_stub; ftrace_shutdown(); /* Restore normal tracing type */ ftrace_tracing_type = FTRACE_TYPE_ENTER; @@ -1612,7 +1612,7 @@ void unregister_ftrace_return(void) } /* Allocate a return stack for newly created task */ -void ftrace_retfunc_init_task(struct task_struct *t) +void ftrace_graph_init_task(struct task_struct *t) { if (atomic_read(&ftrace_retfunc_active)) { t->ret_stack = kmalloc(FTRACE_RETFUNC_DEPTH @@ -1626,7 +1626,7 @@ void ftrace_retfunc_init_task(struct task_struct *t) t->ret_stack = NULL; } -void ftrace_retfunc_exit_task(struct task_struct *t) +void ftrace_graph_exit_task(struct task_struct *t) { struct ftrace_ret_stack *ret_stack = t->ret_stack; diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 8df8fdd69c9..f21ab2c68fd 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -878,15 +878,15 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, ring_buffer_unlock_commit(tr->buffer, event, irq_flags); } -#ifdef CONFIG_FUNCTION_RET_TRACER -static void __trace_function_return(struct trace_array *tr, +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +static void __trace_function_graph(struct trace_array *tr, struct trace_array_cpu *data, - struct ftrace_retfunc *trace, + struct ftrace_graph_ret *trace, unsigned long flags, int pc) { struct ring_buffer_event *event; - struct ftrace_ret_entry *entry; + struct ftrace_graph_entry *entry; unsigned long irq_flags; if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled)))) @@ -1177,8 +1177,8 @@ function_trace_call(unsigned long ip, unsigned long parent_ip) local_irq_restore(flags); } -#ifdef CONFIG_FUNCTION_RET_TRACER -void trace_function_return(struct ftrace_retfunc *trace) +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +void trace_function_graph(struct ftrace_graph_ret *trace) { struct trace_array *tr = &global_trace; struct trace_array_cpu *data; @@ -1193,12 +1193,12 @@ void trace_function_return(struct ftrace_retfunc *trace) disabled = atomic_inc_return(&data->disabled); if (likely(disabled == 1)) { pc = preempt_count(); - __trace_function_return(tr, data, trace, flags, pc); + __trace_function_graph(tr, data, trace, flags, pc); } atomic_dec(&data->disabled); raw_local_irq_restore(flags); } -#endif /* CONFIG_FUNCTION_RET_TRACER */ +#endif /* CONFIG_FUNCTION_GRAPH_TRACER */ static struct ftrace_ops trace_ops __read_mostly = { @@ -2001,7 +2001,7 @@ static enum print_line_t print_trace_fmt(struct trace_iterator *iter) break; } case TRACE_FN_RET: { - return print_return_function(iter); + return print_graph_function(iter); break; } case TRACE_BRANCH: { diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 3abd645e8af..72b5ef86876 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -57,7 +57,7 @@ struct ftrace_entry { }; /* Function return entry */ -struct ftrace_ret_entry { +struct ftrace_graph_entry { struct trace_entry ent; unsigned long ip; unsigned long parent_ip; @@ -264,7 +264,7 @@ extern void __ftrace_bad_type(void); IF_ASSIGN(var, ent, struct trace_boot_call, TRACE_BOOT_CALL);\ IF_ASSIGN(var, ent, struct trace_boot_ret, TRACE_BOOT_RET);\ IF_ASSIGN(var, ent, struct trace_branch, TRACE_BRANCH); \ - IF_ASSIGN(var, ent, struct ftrace_ret_entry, TRACE_FN_RET);\ + IF_ASSIGN(var, ent, struct ftrace_graph_entry, TRACE_FN_RET);\ IF_ASSIGN(var, ent, struct bts_entry, TRACE_BTS);\ __ftrace_bad_type(); \ } while (0) @@ -398,7 +398,7 @@ void trace_function(struct trace_array *tr, unsigned long parent_ip, unsigned long flags, int pc); void -trace_function_return(struct ftrace_retfunc *trace); +trace_function_graph(struct ftrace_graph_ret *trace); void trace_bts(struct trace_array *tr, unsigned long from, @@ -489,11 +489,11 @@ extern int trace_vprintk(unsigned long ip, const char *fmt, va_list args); extern unsigned long trace_flags; /* Standard output formatting function used for function return traces */ -#ifdef CONFIG_FUNCTION_RET_TRACER -extern enum print_line_t print_return_function(struct trace_iterator *iter); +#ifdef CONFIG_FUNCTION_GRAPH_TRACER +extern enum print_line_t print_graph_function(struct trace_iterator *iter); #else static inline enum print_line_t -print_return_function(struct trace_iterator *iter) +print_graph_function(struct trace_iterator *iter) { return TRACE_TYPE_UNHANDLED; } diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c new file mode 100644 index 00000000000..f5bad4624d2 --- /dev/null +++ b/kernel/trace/trace_functions_graph.c @@ -0,0 +1,98 @@ +/* + * + * Function graph tracer. + * Copyright (c) 2008 Frederic Weisbecker + * Mostly borrowed from function tracer which + * is Copyright (c) Steven Rostedt + * + */ +#include +#include +#include +#include + +#include "trace.h" + + +#define TRACE_GRAPH_PRINT_OVERRUN 0x1 +static struct tracer_opt trace_opts[] = { + /* Display overruns or not */ + { TRACER_OPT(overrun, TRACE_GRAPH_PRINT_OVERRUN) }, + { } /* Empty entry */ +}; + +static struct tracer_flags tracer_flags = { + .val = 0, /* Don't display overruns by default */ + .opts = trace_opts +}; + + +static int graph_trace_init(struct trace_array *tr) +{ + int cpu; + for_each_online_cpu(cpu) + tracing_reset(tr, cpu); + + return register_ftrace_graph(&trace_function_graph); +} + +static void graph_trace_reset(struct trace_array *tr) +{ + unregister_ftrace_graph(); +} + + +enum print_line_t +print_graph_function(struct trace_iterator *iter) +{ + struct trace_seq *s = &iter->seq; + struct trace_entry *entry = iter->ent; + struct ftrace_graph_entry *field; + int ret; + + if (entry->type == TRACE_FN_RET) { + trace_assign_type(field, entry); + ret = trace_seq_printf(s, "%pF -> ", (void *)field->parent_ip); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + + ret = seq_print_ip_sym(s, field->ip, + trace_flags & TRACE_ITER_SYM_MASK); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + + ret = trace_seq_printf(s, " (%llu ns)", + field->rettime - field->calltime); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + + if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERRUN) { + ret = trace_seq_printf(s, " (Overruns: %lu)", + field->overrun); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + } + + ret = trace_seq_printf(s, "\n"); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + + return TRACE_TYPE_HANDLED; + } + return TRACE_TYPE_UNHANDLED; +} + +static struct tracer graph_trace __read_mostly = { + .name = "function-graph", + .init = graph_trace_init, + .reset = graph_trace_reset, + .print_line = print_graph_function, + .flags = &tracer_flags, +}; + +static __init int init_graph_trace(void) +{ + return register_tracer(&graph_trace); +} + +device_initcall(init_graph_trace); -- cgit v1.2.3-70-g09d2