From 0eec481e8fb000a209fda9bf8f466aca87dc1150 Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Wed, 15 Oct 2008 14:56:37 +0800 Subject: markers: simplify marker_set_format() current marker_set_format() is complex this patch simplify it, and decrease the overhead of marker_update_probes(). Signed-off-by: Lai Jiangshan Acked-by: Mathieu Desnoyers Signed-off-by: Ingo Molnar --- kernel/marker.c | 71 ++++++++++++++++++++------------------------------------- 1 file changed, 25 insertions(+), 46 deletions(-) diff --git a/kernel/marker.c b/kernel/marker.c index e9c6b2bc9400..75a1a17cd78a 100644 --- a/kernel/marker.c +++ b/kernel/marker.c @@ -64,6 +64,7 @@ struct marker_entry { void *oldptr; int rcu_pending; unsigned char ptype:1; + unsigned char format_allocated:1; char name[0]; /* Contains name'\0'format'\0' */ }; @@ -416,6 +417,7 @@ static struct marker_entry *add_marker(const char *name, const char *format) e->single.probe_private = NULL; e->multi = NULL; e->ptype = 0; + e->format_allocated = 0; e->refcount = 0; e->rcu_pending = 0; hlist_add_head(&e->hlist, head); @@ -447,6 +449,8 @@ static int remove_marker(const char *name) if (e->single.func != __mark_empty_function) return -EBUSY; hlist_del(&e->hlist); + if (e->format_allocated) + kfree(e->format); /* Make sure the call_rcu has been executed */ if (e->rcu_pending) rcu_barrier_sched(); @@ -457,57 +461,34 @@ static int remove_marker(const char *name) /* * Set the mark_entry format to the format found in the element. */ -static int marker_set_format(struct marker_entry **entry, const char *format) +static int marker_set_format(struct marker_entry *entry, const char *format) { - struct marker_entry *e; - size_t name_len = strlen((*entry)->name) + 1; - size_t format_len = strlen(format) + 1; - - - e = kmalloc(sizeof(struct marker_entry) + name_len + format_len, - GFP_KERNEL); - if (!e) + entry->format = kstrdup(format, GFP_KERNEL); + if (!entry->format) return -ENOMEM; - memcpy(&e->name[0], (*entry)->name, name_len); - e->format = &e->name[name_len]; - memcpy(e->format, format, format_len); - if (strcmp(e->format, MARK_NOARGS) == 0) - e->call = marker_probe_cb_noarg; - else - e->call = marker_probe_cb; - e->single = (*entry)->single; - e->multi = (*entry)->multi; - e->ptype = (*entry)->ptype; - e->refcount = (*entry)->refcount; - e->rcu_pending = 0; - hlist_add_before(&e->hlist, &(*entry)->hlist); - hlist_del(&(*entry)->hlist); - /* Make sure the call_rcu has been executed */ - if ((*entry)->rcu_pending) - rcu_barrier_sched(); - kfree(*entry); - *entry = e; + entry->format_allocated = 1; + trace_mark(core_marker_format, "name %s format %s", - e->name, e->format); + entry->name, entry->format); return 0; } /* * Sets the probe callback corresponding to one marker. */ -static int set_marker(struct marker_entry **entry, struct marker *elem, +static int set_marker(struct marker_entry *entry, struct marker *elem, int active) { int ret; - WARN_ON(strcmp((*entry)->name, elem->name) != 0); + WARN_ON(strcmp(entry->name, elem->name) != 0); - if ((*entry)->format) { - if (strcmp((*entry)->format, elem->format) != 0) { + if (entry->format) { + if (strcmp(entry->format, elem->format) != 0) { printk(KERN_NOTICE "Format mismatch for probe %s " "(%s), marker (%s)\n", - (*entry)->name, - (*entry)->format, + entry->name, + entry->format, elem->format); return -EPERM; } @@ -523,34 +504,33 @@ static int set_marker(struct marker_entry **entry, struct marker *elem, * pass from a "safe" callback (with argument) to an "unsafe" * callback (does not set arguments). */ - elem->call = (*entry)->call; + elem->call = entry->call; /* * Sanity check : * We only update the single probe private data when the ptr is * set to a _non_ single probe! (0 -> 1 and N -> 1, N != 1) */ WARN_ON(elem->single.func != __mark_empty_function - && elem->single.probe_private - != (*entry)->single.probe_private && - !elem->ptype); - elem->single.probe_private = (*entry)->single.probe_private; + && elem->single.probe_private != entry->single.probe_private + && !elem->ptype); + elem->single.probe_private = entry->single.probe_private; /* * Make sure the private data is valid when we update the * single probe ptr. */ smp_wmb(); - elem->single.func = (*entry)->single.func; + elem->single.func = entry->single.func; /* * We also make sure that the new probe callbacks array is consistent * before setting a pointer to it. */ - rcu_assign_pointer(elem->multi, (*entry)->multi); + rcu_assign_pointer(elem->multi, entry->multi); /* * Update the function or multi probe array pointer before setting the * ptype. */ smp_wmb(); - elem->ptype = (*entry)->ptype; + elem->ptype = entry->ptype; elem->state = active; return 0; @@ -594,8 +574,7 @@ void marker_update_probe_range(struct marker *begin, for (iter = begin; iter < end; iter++) { mark_entry = get_marker(iter->name); if (mark_entry) { - set_marker(&mark_entry, iter, - !!mark_entry->refcount); + set_marker(mark_entry, iter, !!mark_entry->refcount); /* * ignore error, continue */ @@ -657,7 +636,7 @@ int marker_probe_register(const char *name, const char *format, ret = PTR_ERR(entry); } else if (format) { if (!entry->format) - ret = marker_set_format(&entry, format); + ret = marker_set_format(entry, format); else if (strcmp(entry->format, format)) ret = -EPERM; } -- cgit v1.2.3-59-g8ed1b From 505e371da195fad20cb8aaf45407a2849774d6d0 Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Wed, 15 Oct 2008 14:56:42 +0800 Subject: markers: remove exported symbol marker_probe_cb_noarg() marker_probe_cb_noarg() should not be seen by outer code. this patch remove it. Signed-off-by: Lai Jiangshan Acked-by: Mathieu Desnoyers Signed-off-by: Ingo Molnar --- include/linux/marker.h | 2 -- kernel/marker.c | 3 +-- 2 files changed, 1 insertion(+), 4 deletions(-) diff --git a/include/linux/marker.h b/include/linux/marker.h index 889196c7fbb1..4cf45472d9f5 100644 --- a/include/linux/marker.h +++ b/include/linux/marker.h @@ -136,8 +136,6 @@ extern marker_probe_func __mark_empty_function; extern void marker_probe_cb(const struct marker *mdata, void *call_private, ...); -extern void marker_probe_cb_noarg(const struct marker *mdata, - void *call_private, ...); /* * Connect a probe to a marker. diff --git a/kernel/marker.c b/kernel/marker.c index 75a1a17cd78a..23192646555f 100644 --- a/kernel/marker.c +++ b/kernel/marker.c @@ -158,7 +158,7 @@ EXPORT_SYMBOL_GPL(marker_probe_cb); * * Should be connected to markers "MARK_NOARGS". */ -void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...) +static void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...) { va_list args; /* not initialized */ char ptype; @@ -198,7 +198,6 @@ void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...) } rcu_read_unlock_sched(); } -EXPORT_SYMBOL_GPL(marker_probe_cb_noarg); static void free_old_closure(struct rcu_head *head) { -- cgit v1.2.3-59-g8ed1b From 4de62748e69c31fc4fd5bc43b73e9cf60a17ec53 Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Wed, 15 Oct 2008 14:56:47 +0800 Subject: markers: let marker_table be close to its comments marker_table is defined far from its comments, this fix make cleanup for it. Signed-off-by: Lai Jiangshan Acked-by: Mathieu Desnoyers Signed-off-by: Ingo Molnar --- kernel/marker.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/kernel/marker.c b/kernel/marker.c index 23192646555f..0f2a944329d3 100644 --- a/kernel/marker.c +++ b/kernel/marker.c @@ -43,6 +43,7 @@ static DEFINE_MUTEX(markers_mutex); */ #define MARKER_HASH_BITS 6 #define MARKER_TABLE_SIZE (1 << MARKER_HASH_BITS) +static struct hlist_head marker_table[MARKER_TABLE_SIZE]; /* * Note about RCU : @@ -68,8 +69,6 @@ struct marker_entry { char name[0]; /* Contains name'\0'format'\0' */ }; -static struct hlist_head marker_table[MARKER_TABLE_SIZE]; - /** * __mark_empty_function - Empty probe callback * @probe_private: probe private data -- cgit v1.2.3-59-g8ed1b From 5d9881ea1440f046ee851bbaa2a2962543336a11 Mon Sep 17 00:00:00 2001 From: Zhaolei Date: Wed, 22 Oct 2008 11:38:01 +0800 Subject: markers: break the redundant loop in kernel/marker.c Impact: cleanup, no functionality changed Because e->name is unique in list, we don't need to continue the iteration after matched. Signed-off-by: Zhao Lei Acked-by: Mathieu Desnoyers Signed-off-by: Ingo Molnar --- kernel/marker.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/kernel/marker.c b/kernel/marker.c index 0f2a944329d3..2898b647d415 100644 --- a/kernel/marker.c +++ b/kernel/marker.c @@ -825,8 +825,6 @@ void *marker_get_private_data(const char *name, marker_probe_func *probe, if (!e->ptype) { if (num == 0 && e->single.func == probe) return e->single.probe_private; - else - break; } else { struct marker_probe_closure *closure; int match = 0; @@ -838,6 +836,7 @@ void *marker_get_private_data(const char *name, marker_probe_func *probe, return closure[i].probe_private; } } + break; } } return ERR_PTR(-ENOENT); -- cgit v1.2.3-59-g8ed1b From fd3fdf11d3c649769e02459c5f1b8081a15e9007 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Fri, 24 Oct 2008 20:08:11 +0300 Subject: trace: add the MMIO-tracer to the tracer menu, cleanup Impact: cleanup We can remove MMIOTRACE_HOOKS and replace it with just MMIOTRACE. MMIOTRACE_HOOKS is a remnant from the time when I thought that something else could also use the kmmio facilities. Signed-off-by: Pekka Paalanen Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/Kconfig.debug | 4 ---- arch/x86/mm/Makefile | 3 +-- arch/x86/mm/fault.c | 2 +- 3 files changed, 2 insertions(+), 7 deletions(-) diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug index 2a3dfbd5e677..fa013f529b74 100644 --- a/arch/x86/Kconfig.debug +++ b/arch/x86/Kconfig.debug @@ -186,14 +186,10 @@ config IOMMU_LEAK Add a simple leak tracer to the IOMMU code. This is useful when you are debugging a buggy device driver that leaks IOMMU mappings. -config MMIOTRACE_HOOKS - bool - config MMIOTRACE bool "Memory mapped IO tracing" depends on DEBUG_KERNEL && PCI select TRACING - select MMIOTRACE_HOOKS help Mmiotrace traces Memory Mapped I/O access and is meant for debugging and reverse engineering. It is called from the ioremap diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile index 59f89b434b45..0a21b7aab9dc 100644 --- a/arch/x86/mm/Makefile +++ b/arch/x86/mm/Makefile @@ -8,9 +8,8 @@ obj-$(CONFIG_X86_PTDUMP) += dump_pagetables.o obj-$(CONFIG_HIGHMEM) += highmem_32.o -obj-$(CONFIG_MMIOTRACE_HOOKS) += kmmio.o obj-$(CONFIG_MMIOTRACE) += mmiotrace.o -mmiotrace-y := pf_in.o mmio-mod.o +mmiotrace-y := kmmio.o pf_in.o mmio-mod.o obj-$(CONFIG_MMIOTRACE_TEST) += testmmiotrace.o obj-$(CONFIG_NUMA) += numa_$(BITS).o diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c index 31e8730fa246..4152d3c3b138 100644 --- a/arch/x86/mm/fault.c +++ b/arch/x86/mm/fault.c @@ -53,7 +53,7 @@ static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr) { -#ifdef CONFIG_MMIOTRACE_HOOKS +#ifdef CONFIG_MMIOTRACE if (unlikely(is_kmmio_active())) if (kmmio_handler(regs, addr) == 1) return -1; -- cgit v1.2.3-59-g8ed1b 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/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 | 15 ++++++ arch/x86/kernel/ftrace.c | 107 +++++++++++++++++++++++++++++++++++++- include/linux/hardirq.h | 15 +++++- 7 files changed, 154 insertions(+), 3 deletions(-) diff --git a/arch/arm/include/asm/ftrace.h b/arch/arm/include/asm/ftrace.h index 39c8bc1a006a..d4c24a7a9280 100644 --- a/arch/arm/include/asm/ftrace.h +++ b/arch/arm/include/asm/ftrace.h @@ -1,6 +1,11 @@ #ifndef _ASM_ARM_FTRACE #define _ASM_ARM_FTRACE +#ifndef __ASSEMBLY__ +#define ftrace_nmi_enter() do { } while (0) +#define ftrace_nmi_exit() do { } while (0) +#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 b298f7a631e6..7652755dc000 100644 --- a/arch/powerpc/include/asm/ftrace.h +++ b/arch/powerpc/include/asm/ftrace.h @@ -1,6 +1,11 @@ #ifndef _ASM_POWERPC_FTRACE #define _ASM_POWERPC_FTRACE +#ifndef __ASSEMBLY__ +#define ftrace_nmi_enter() do { } while (0) +#define ftrace_nmi_exit() do { } while (0) +#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 3aed362c9463..cdf2cb0b9ffe 100644 --- a/arch/sh/include/asm/ftrace.h +++ b/arch/sh/include/asm/ftrace.h @@ -1,6 +1,11 @@ #ifndef __ASM_SH_FTRACE_H #define __ASM_SH_FTRACE_H +#ifndef __ASSEMBLY__ +#define ftrace_nmi_enter() do { } while (0) +#define ftrace_nmi_exit() do { } while (0) +#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 d27716cd38c1..33a95feeb137 100644 --- a/arch/sparc/include/asm/ftrace.h +++ b/arch/sparc/include/asm/ftrace.h @@ -1,6 +1,11 @@ #ifndef _ASM_SPARC64_FTRACE #define _ASM_SPARC64_FTRACE +#ifndef __ASSEMBLY__ +#define ftrace_nmi_enter() do { } while (0) +#define ftrace_nmi_exit() do { } while (0) +#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 9e8bc29b8b17..f2ed6b704a75 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 */ diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 50ea0ac8c9bf..fe5f859130b5 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -56,6 +56,111 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr) return calc.code; } +/* + * Modifying code must take extra care. On an SMP machine, if + * the code being modified is also being executed on another CPU + * that CPU will have undefined results and possibly take a GPF. + * We use kstop_machine to stop other CPUS from exectuing code. + * But this does not stop NMIs from happening. We still need + * to protect against that. We separate out the modification of + * the code to take care of this. + * + * Two buffers are added: An IP buffer and a "code" buffer. + * + * 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, the first thing it does is to call + * "ftrace_nmi_enter". This will check if the flag is set to write + * and if it is, it will write what is in the IP and "code" buffers. + * + * The trick is, it does not matter if everyone is writing the same + * content to the code location. Also, if a CPU is executing code + * it is OK to write to that code location if the contents being written + * 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 void ftrace_mod_code(void) +{ + /* + * Yes, more than one CPU process can be writing to mod_code_status. + * (and the code itself) + * But if one were to fail, then they all should, and if one were + * to succeed, then they all should. + */ + mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode, + MCOUNT_INSN_SIZE); + +} + +void ftrace_nmi_enter(void) +{ + atomic_inc(&in_nmi); + /* Must have in_nmi seen before reading write flag */ + smp_mb(); + if (mod_code_write) + ftrace_mod_code(); +} + +void ftrace_nmi_exit(void) +{ + /* Finish all executions before clearing in_nmi */ + smp_wmb(); + atomic_dec(&in_nmi); +} + +static void wait_for_nmi(void) +{ + while (atomic_read(&in_nmi)) + cpu_relax(); +} + +static int +do_ftrace_mod_code(unsigned long ip, void *new_code) +{ + mod_code_ip = (void *)ip; + mod_code_newcode = new_code; + + /* The buffers need to be visible before we let NMIs write them */ + smp_wmb(); + + mod_code_write = 1; + + /* Make sure write bit is visible before we wait on NMIs */ + smp_mb(); + + wait_for_nmi(); + + /* Make sure all running NMIs have finished before we write the code */ + smp_mb(); + + ftrace_mod_code(); + + /* Make sure the write happens before clearing the bit */ + smp_wmb(); + + mod_code_write = 0; + + /* make sure NMIs see the cleared bit */ + smp_mb(); + + wait_for_nmi(); + + return mod_code_status; +} + + int ftrace_modify_code(unsigned long ip, unsigned char *old_code, unsigned char *new_code) @@ -81,7 +186,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, return -EINVAL; /* replace the text with the new text */ - if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE)) + if (do_ftrace_mod_code(ip, new_code)) return -EPERM; sync_core(); diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h index 181006cc94a0..0087cb43becf 100644 --- a/include/linux/hardirq.h +++ b/include/linux/hardirq.h @@ -5,6 +5,7 @@ #include #include #include +#include #include /* @@ -161,7 +162,17 @@ extern void irq_enter(void); */ extern void irq_exit(void); -#define nmi_enter() do { lockdep_off(); __irq_enter(); } while (0) -#define nmi_exit() do { __irq_exit(); lockdep_on(); } while (0) +#define nmi_enter() \ + do { \ + ftrace_nmi_enter(); \ + lockdep_off(); \ + __irq_enter(); \ + } while (0) +#define nmi_exit() \ + do { \ + __irq_exit(); \ + lockdep_on(); \ + ftrace_nmi_exit(); \ + } while (0) #endif /* LINUX_HARDIRQ_H */ -- cgit v1.2.3-59-g8ed1b From b807c3d0f8e39ed7cbbbe6da162650e305e8de15 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 30 Oct 2008 16:08:33 -0400 Subject: ftrace: nmi update statistics Impact: add more debug info to /debugfs/tracing/dyn_ftrace_total_info This patch adds dynamic ftrace NMI update statistics to the /debugfs/tracing/dyn_ftrace_total_info stat file. Signed-off-by: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Cc: Linus Torvalds Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- arch/x86/kernel/ftrace.c | 26 ++++++++++++++++++++++++-- kernel/trace/trace.c | 31 ++++++++++++++++++++++++------- 2 files changed, 48 insertions(+), 9 deletions(-) diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index fe5f859130b5..6685b0fc1b44 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -91,6 +91,19 @@ static int mod_code_write; static void *mod_code_ip; static void *mod_code_newcode; +static int nmi_wait_count; +static atomic_t nmi_update_count; + +int ftrace_arch_read_dyn_info(char *buf, int size) +{ + int r; + + r = snprintf(buf, size, "%u %u", + nmi_wait_count, + atomic_read(&nmi_update_count)); + return r; +} + static void ftrace_mod_code(void) { /* @@ -109,8 +122,10 @@ void ftrace_nmi_enter(void) atomic_inc(&in_nmi); /* Must have in_nmi seen before reading write flag */ smp_mb(); - if (mod_code_write) + if (mod_code_write) { ftrace_mod_code(); + atomic_inc(&nmi_update_count); + } } void ftrace_nmi_exit(void) @@ -122,8 +137,15 @@ void ftrace_nmi_exit(void) static void wait_for_nmi(void) { - while (atomic_read(&in_nmi)) + int waited = 0; + + while (atomic_read(&in_nmi)) { + waited = 1; cpu_relax(); + } + + if (waited) + nmi_wait_count++; } static int diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index a610ca771558..bc36febc0771 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2815,22 +2815,39 @@ 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; +} + static ssize_t -tracing_read_long(struct file *filp, char __user *ubuf, +tracing_read_dyn_info(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) { unsigned long *p = filp->private_data; - char buf[64]; + char *buf = ftrace_dyn_info_buffer; int r; - r = sprintf(buf, "%ld\n", *p); + mutex_lock(&dyn_info_mutex); + r = sprintf(buf, "%ld ", *p); - return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); + r += ftrace_arch_read_dyn_info(buf+r, DYN_INFO_BUF_SIZE-r); + buf[r++] = '\n'; + + r = simple_read_from_buffer(ubuf, cnt, ppos, buf, r); + + mutex_unlock(&dyn_info_mutex); + + return r; } -static struct file_operations tracing_read_long_fops = { +static struct file_operations tracing_dyn_info_fops = { .open = tracing_open_generic, - .read = tracing_read_long, + .read = tracing_read_dyn_info, }; #endif @@ -2939,7 +2956,7 @@ static __init int tracer_init_debugfs(void) #ifdef CONFIG_DYNAMIC_FTRACE entry = debugfs_create_file("dyn_ftrace_total_info", 0444, d_tracer, &ftrace_update_tot_cnt, - &tracing_read_long_fops); + &tracing_dyn_info_fops); if (!entry) pr_warning("Could not create debugfs " "'dyn_ftrace_total_info' entry\n"); -- cgit v1.2.3-59-g8ed1b 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(-) diff --git a/arch/arm/include/asm/ftrace.h b/arch/arm/include/asm/ftrace.h index d4c24a7a9280..3f3a1d1508ea 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 7652755dc000..1cd72700fbc0 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 cdf2cb0b9ffe..31ada0370cb6 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 33a95feeb137..62055ac0496e 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 f2ed6b704a75..a23468194b8c 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 6685b0fc1b44..69149337f2fe 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 703eb53cfa2b..22240dfe912e 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 bc36febc0771..7f86067d760c 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 v1.2.3-59-g8ed1b From c2c80529460095035752bf0ecc1af82c1e0f6e0f Mon Sep 17 00:00:00 2001 From: Al Viro Date: Fri, 31 Oct 2008 19:50:41 +0000 Subject: tracing, alpha: undefined reference to `save_stack_trace' Impact: build fix on !stacktrace architectures only select STACKTRACE on architectures that have STACKTRACE_SUPPORT ... since we also need to ifdef out the guts of ftrace_trace_stack(). We also want to disallow setting TRACE_ITER_STACKTRACE in trace_flags on such configs, but that can wait. Signed-off-by: Al Viro Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/Kconfig | 2 +- kernel/trace/trace.c | 2 ++ 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index b58f43bec363..33dbefd471e8 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -25,7 +25,7 @@ config TRACING bool select DEBUG_FS select RING_BUFFER - select STACKTRACE + select STACKTRACE if STACKTRACE_SUPPORT select TRACEPOINTS select NOP_TRACER diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 8a499e2adaec..85bee775a03e 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -705,6 +705,7 @@ static void ftrace_trace_stack(struct trace_array *tr, unsigned long flags, int skip, int pc) { +#ifdef CONFIG_STACKTRACE struct ring_buffer_event *event; struct stack_entry *entry; struct stack_trace trace; @@ -730,6 +731,7 @@ static void ftrace_trace_stack(struct trace_array *tr, save_stack_trace(&trace); ring_buffer_unlock_commit(tr->buffer, event, irq_flags); +#endif } void __trace_stack(struct trace_array *tr, -- cgit v1.2.3-59-g8ed1b From 19dba33c43a2f0f2aa727ae075ec3b11330775ef Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Tue, 28 Oct 2008 10:51:49 +0800 Subject: tracepoint: simplification for tracepoints using RCU Impact: simplify implementation Now, unused memory is handled by struct tp_probes. old code use these three field to handle unused memory. struct tracepoint_entry { ... struct rcu_head rcu; void *oldptr; unsigned char rcu_pending:1; ... }; in this way, unused memory is handled by struct tracepoint_entry. it bring reenter bug(it was fixed) and tracepoint.c is filled full of ".*rcu.*" code statements. this patch removes all these. and: rcu_barrier_sched() is removed. Do not need regain tracepoints_mutex after tracepoint_update_probes() several little cleanup. Signed-off-by: Lai Jiangshan Acked-by: Mathieu Desnoyers Signed-off-by: Ingo Molnar --- kernel/tracepoint.c | 111 ++++++++++++++++++---------------------------------- 1 file changed, 37 insertions(+), 74 deletions(-) diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index af8c85664882..3e22867184e3 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -43,6 +43,7 @@ static DEFINE_MUTEX(tracepoints_mutex); */ #define TRACEPOINT_HASH_BITS 6 #define TRACEPOINT_TABLE_SIZE (1 << TRACEPOINT_HASH_BITS) +static struct hlist_head tracepoint_table[TRACEPOINT_TABLE_SIZE]; /* * Note about RCU : @@ -54,40 +55,40 @@ struct tracepoint_entry { struct hlist_node hlist; void **funcs; int refcount; /* Number of times armed. 0 if disarmed. */ - struct rcu_head rcu; - void *oldptr; - unsigned char rcu_pending:1; char name[0]; }; -static struct hlist_head tracepoint_table[TRACEPOINT_TABLE_SIZE]; +struct tp_probes { + struct rcu_head rcu; + void *probes[0]; +}; -static void free_old_closure(struct rcu_head *head) +static inline void *allocate_probes(int count) { - struct tracepoint_entry *entry = container_of(head, - struct tracepoint_entry, rcu); - kfree(entry->oldptr); - /* Make sure we free the data before setting the pending flag to 0 */ - smp_wmb(); - entry->rcu_pending = 0; + struct tp_probes *p = kmalloc(count * sizeof(void *) + + sizeof(struct tp_probes), GFP_KERNEL); + return p == NULL ? NULL : p->probes; } -static void tracepoint_entry_free_old(struct tracepoint_entry *entry, void *old) +static void rcu_free_old_probes(struct rcu_head *head) { - if (!old) - return; - entry->oldptr = old; - entry->rcu_pending = 1; - /* write rcu_pending before calling the RCU callback */ - smp_wmb(); - call_rcu_sched(&entry->rcu, free_old_closure); + kfree(container_of(head, struct tp_probes, rcu)); +} + +static inline void release_probes(void *old) +{ + if (old) { + struct tp_probes *tp_probes = container_of(old, + struct tp_probes, probes[0]); + call_rcu(&tp_probes->rcu, rcu_free_old_probes); + } } static void debug_print_probes(struct tracepoint_entry *entry) { int i; - if (!tracepoint_debug) + if (!tracepoint_debug || !entry->funcs) return; for (i = 0; entry->funcs[i]; i++) @@ -111,12 +112,13 @@ tracepoint_entry_add_probe(struct tracepoint_entry *entry, void *probe) return ERR_PTR(-EEXIST); } /* + 2 : one for new probe, one for NULL func */ - new = kzalloc((nr_probes + 2) * sizeof(void *), GFP_KERNEL); + new = allocate_probes(nr_probes + 2); if (new == NULL) return ERR_PTR(-ENOMEM); if (old) memcpy(new, old, nr_probes * sizeof(void *)); new[nr_probes] = probe; + new[nr_probes + 1] = NULL; entry->refcount = nr_probes + 1; entry->funcs = new; debug_print_probes(entry); @@ -132,7 +134,7 @@ tracepoint_entry_remove_probe(struct tracepoint_entry *entry, void *probe) old = entry->funcs; if (!old) - return NULL; + return ERR_PTR(-ENOENT); debug_print_probes(entry); /* (N -> M), (N > 1, M >= 0) probes */ @@ -151,13 +153,13 @@ tracepoint_entry_remove_probe(struct tracepoint_entry *entry, void *probe) int j = 0; /* N -> M, (N > 1, M > 0) */ /* + 1 for NULL */ - new = kzalloc((nr_probes - nr_del + 1) - * sizeof(void *), GFP_KERNEL); + new = allocate_probes(nr_probes - nr_del + 1); if (new == NULL) return ERR_PTR(-ENOMEM); for (i = 0; old[i]; i++) if ((probe && old[i] != probe)) new[j++] = old[i]; + new[nr_probes - nr_del] = NULL; entry->refcount = nr_probes - nr_del; entry->funcs = new; } @@ -215,7 +217,6 @@ static struct tracepoint_entry *add_tracepoint(const char *name) memcpy(&e->name[0], name, name_len); e->funcs = NULL; e->refcount = 0; - e->rcu_pending = 0; hlist_add_head(&e->hlist, head); return e; } @@ -224,32 +225,10 @@ static struct tracepoint_entry *add_tracepoint(const char *name) * Remove the tracepoint from the tracepoint hash table. Must be called with * mutex_lock held. */ -static int remove_tracepoint(const char *name) +static inline void remove_tracepoint(struct tracepoint_entry *e) { - struct hlist_head *head; - struct hlist_node *node; - struct tracepoint_entry *e; - int found = 0; - size_t len = strlen(name) + 1; - u32 hash = jhash(name, len-1, 0); - - head = &tracepoint_table[hash & (TRACEPOINT_TABLE_SIZE - 1)]; - hlist_for_each_entry(e, node, head, hlist) { - if (!strcmp(name, e->name)) { - found = 1; - break; - } - } - if (!found) - return -ENOENT; - if (e->refcount) - return -EBUSY; hlist_del(&e->hlist); - /* Make sure the call_rcu_sched has been executed */ - if (e->rcu_pending) - rcu_barrier_sched(); kfree(e); - return 0; } /* @@ -343,25 +322,17 @@ int tracepoint_probe_register(const char *name, void *probe) goto end; } } - /* - * If we detect that a call_rcu_sched is pending for this tracepoint, - * make sure it's executed now. - */ - if (entry->rcu_pending) - rcu_barrier_sched(); old = tracepoint_entry_add_probe(entry, probe); if (IS_ERR(old)) { + if (!entry->refcount) + remove_tracepoint(entry); ret = PTR_ERR(old); goto end; } mutex_unlock(&tracepoints_mutex); tracepoint_update_probes(); /* may update entry */ - mutex_lock(&tracepoints_mutex); - entry = get_tracepoint(name); - WARN_ON(!entry); - if (entry->rcu_pending) - rcu_barrier_sched(); - tracepoint_entry_free_old(entry, old); + release_probes(old); + return 0; end: mutex_unlock(&tracepoints_mutex); return ret; @@ -388,25 +359,17 @@ int tracepoint_probe_unregister(const char *name, void *probe) entry = get_tracepoint(name); if (!entry) goto end; - if (entry->rcu_pending) - rcu_barrier_sched(); old = tracepoint_entry_remove_probe(entry, probe); - if (!old) { - printk(KERN_WARNING "Warning: Trying to unregister a probe" - "that doesn't exist\n"); + if (IS_ERR(old)) { + ret = PTR_ERR(old); goto end; } + if (!entry->refcount) + remove_tracepoint(entry); mutex_unlock(&tracepoints_mutex); tracepoint_update_probes(); /* may update entry */ - mutex_lock(&tracepoints_mutex); - entry = get_tracepoint(name); - if (!entry) - goto end; - if (entry->rcu_pending) - rcu_barrier_sched(); - tracepoint_entry_free_old(entry, old); - remove_tracepoint(name); /* Ignore busy error message */ - ret = 0; + release_probes(old); + return 0; end: mutex_unlock(&tracepoints_mutex); return ret; -- cgit v1.2.3-59-g8ed1b From 127cafbb276266b1b8da967bfe25a062ab1d42ab Mon Sep 17 00:00:00 2001 From: Lai Jiangshan Date: Tue, 28 Oct 2008 10:51:53 +0800 Subject: tracepoint: introduce *_noupdate APIs. Impact: add new tracepoint APIs to allow the batched registration of probes new APIs separate tracepoint_probe_register(), tracepoint_probe_unregister() into 2 steps. The first step of them is just update tracepoint_entry, not connect or disconnect. this patch introduces tracepoint_probe_update_all() for update all. these APIs are very useful for registering lots of probes but just updating once. Another very important thing is that *_noupdate APIs do not require module_mutex. Signed-off-by: Lai Jiangshan Acked-by: Mathieu Desnoyers Signed-off-by: Ingo Molnar --- include/linux/tracepoint.h | 4 ++ kernel/tracepoint.c | 170 +++++++++++++++++++++++++++++++++++---------- 2 files changed, 136 insertions(+), 38 deletions(-) diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h index c5bb39c7a770..63064e9403f2 100644 --- a/include/linux/tracepoint.h +++ b/include/linux/tracepoint.h @@ -112,6 +112,10 @@ extern int tracepoint_probe_register(const char *name, void *probe); */ extern int tracepoint_probe_unregister(const char *name, void *probe); +extern int tracepoint_probe_register_noupdate(const char *name, void *probe); +extern int tracepoint_probe_unregister_noupdate(const char *name, void *probe); +extern void tracepoint_probe_update_all(void); + struct tracepoint_iter { struct module *module; struct tracepoint *tracepoint; diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c index 3e22867184e3..e96590f17de1 100644 --- a/kernel/tracepoint.c +++ b/kernel/tracepoint.c @@ -59,7 +59,10 @@ struct tracepoint_entry { }; struct tp_probes { - struct rcu_head rcu; + union { + struct rcu_head rcu; + struct list_head list; + } u; void *probes[0]; }; @@ -72,7 +75,7 @@ static inline void *allocate_probes(int count) static void rcu_free_old_probes(struct rcu_head *head) { - kfree(container_of(head, struct tp_probes, rcu)); + kfree(container_of(head, struct tp_probes, u.rcu)); } static inline void release_probes(void *old) @@ -80,7 +83,7 @@ static inline void release_probes(void *old) if (old) { struct tp_probes *tp_probes = container_of(old, struct tp_probes, probes[0]); - call_rcu(&tp_probes->rcu, rcu_free_old_probes); + call_rcu_sched(&tp_probes->u.rcu, rcu_free_old_probes); } } @@ -299,6 +302,23 @@ static void tracepoint_update_probes(void) module_update_tracepoints(); } +static void *tracepoint_add_probe(const char *name, void *probe) +{ + struct tracepoint_entry *entry; + void *old; + + entry = get_tracepoint(name); + if (!entry) { + entry = add_tracepoint(name); + if (IS_ERR(entry)) + return entry; + } + old = tracepoint_entry_add_probe(entry, probe); + if (IS_ERR(old) && !entry->refcount) + remove_tracepoint(entry); + return old; +} + /** * tracepoint_probe_register - Connect a probe to a tracepoint * @name: tracepoint name @@ -309,36 +329,36 @@ static void tracepoint_update_probes(void) */ int tracepoint_probe_register(const char *name, void *probe) { - struct tracepoint_entry *entry; - int ret = 0; void *old; mutex_lock(&tracepoints_mutex); - entry = get_tracepoint(name); - if (!entry) { - entry = add_tracepoint(name); - if (IS_ERR(entry)) { - ret = PTR_ERR(entry); - goto end; - } - } - old = tracepoint_entry_add_probe(entry, probe); - if (IS_ERR(old)) { - if (!entry->refcount) - remove_tracepoint(entry); - ret = PTR_ERR(old); - goto end; - } + old = tracepoint_add_probe(name, probe); mutex_unlock(&tracepoints_mutex); + if (IS_ERR(old)) + return PTR_ERR(old); + tracepoint_update_probes(); /* may update entry */ release_probes(old); return 0; -end: - mutex_unlock(&tracepoints_mutex); - return ret; } EXPORT_SYMBOL_GPL(tracepoint_probe_register); +static void *tracepoint_remove_probe(const char *name, void *probe) +{ + struct tracepoint_entry *entry; + void *old; + + entry = get_tracepoint(name); + if (!entry) + return ERR_PTR(-ENOENT); + old = tracepoint_entry_remove_probe(entry, probe); + if (IS_ERR(old)) + return old; + if (!entry->refcount) + remove_tracepoint(entry); + return old; +} + /** * tracepoint_probe_unregister - Disconnect a probe from a tracepoint * @name: tracepoint name @@ -351,31 +371,105 @@ EXPORT_SYMBOL_GPL(tracepoint_probe_register); */ int tracepoint_probe_unregister(const char *name, void *probe) { - struct tracepoint_entry *entry; void *old; - int ret = -ENOENT; mutex_lock(&tracepoints_mutex); - entry = get_tracepoint(name); - if (!entry) - goto end; - old = tracepoint_entry_remove_probe(entry, probe); - if (IS_ERR(old)) { - ret = PTR_ERR(old); - goto end; - } - if (!entry->refcount) - remove_tracepoint(entry); + old = tracepoint_remove_probe(name, probe); mutex_unlock(&tracepoints_mutex); + if (IS_ERR(old)) + return PTR_ERR(old); + tracepoint_update_probes(); /* may update entry */ release_probes(old); return 0; -end: - mutex_unlock(&tracepoints_mutex); - return ret; } EXPORT_SYMBOL_GPL(tracepoint_probe_unregister); +static LIST_HEAD(old_probes); +static int need_update; + +static void tracepoint_add_old_probes(void *old) +{ + need_update = 1; + if (old) { + struct tp_probes *tp_probes = container_of(old, + struct tp_probes, probes[0]); + list_add(&tp_probes->u.list, &old_probes); + } +} + +/** + * tracepoint_probe_register_noupdate - register a probe but not connect + * @name: tracepoint name + * @probe: probe handler + * + * caller must call tracepoint_probe_update_all() + */ +int tracepoint_probe_register_noupdate(const char *name, void *probe) +{ + void *old; + + mutex_lock(&tracepoints_mutex); + old = tracepoint_add_probe(name, probe); + if (IS_ERR(old)) { + mutex_unlock(&tracepoints_mutex); + return PTR_ERR(old); + } + tracepoint_add_old_probes(old); + mutex_unlock(&tracepoints_mutex); + return 0; +} +EXPORT_SYMBOL_GPL(tracepoint_probe_register_noupdate); + +/** + * tracepoint_probe_unregister_noupdate - remove a probe but not disconnect + * @name: tracepoint name + * @probe: probe function pointer + * + * caller must call tracepoint_probe_update_all() + */ +int tracepoint_probe_unregister_noupdate(const char *name, void *probe) +{ + void *old; + + mutex_lock(&tracepoints_mutex); + old = tracepoint_remove_probe(name, probe); + if (IS_ERR(old)) { + mutex_unlock(&tracepoints_mutex); + return PTR_ERR(old); + } + tracepoint_add_old_probes(old); + mutex_unlock(&tracepoints_mutex); + return 0; +} +EXPORT_SYMBOL_GPL(tracepoint_probe_unregister_noupdate); + +/** + * tracepoint_probe_update_all - update tracepoints + */ +void tracepoint_probe_update_all(void) +{ + LIST_HEAD(release_probes); + struct tp_probes *pos, *next; + + mutex_lock(&tracepoints_mutex); + if (!need_update) { + mutex_unlock(&tracepoints_mutex); + return; + } + if (!list_empty(&old_probes)) + list_replace_init(&old_probes, &release_probes); + need_update = 0; + mutex_unlock(&tracepoints_mutex); + + tracepoint_update_probes(); + list_for_each_entry_safe(pos, next, &release_probes, u.list) { + list_del(&pos->u.list); + call_rcu_sched(&pos->u.rcu, rcu_free_old_probes); + } +} +EXPORT_SYMBOL_GPL(tracepoint_probe_update_all); + /** * tracepoint_get_iter_range - Get a next tracepoint iterator given a range. * @tracepoint: current tracepoints (in), next tracepoint (out) -- cgit v1.2.3-59-g8ed1b