From 5168ae50a66e3ff7184c2b16d661bd6d70367e50 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 3 Jun 2010 09:36:50 -0400 Subject: tracing: Remove ftrace_preempt_disable/enable The ftrace_preempt_disable/enable functions were to address a recursive race caused by the function tracer. The function tracer traces all functions which makes it easily susceptible to recursion. One area was preempt_enable(). This would call the scheduler and the schedulre would call the function tracer and loop. (So was it thought). The ftrace_preempt_disable/enable was made to protect against recursion inside the scheduler by storing the NEED_RESCHED flag. If it was set before the ftrace_preempt_disable() it would not call schedule on ftrace_preempt_enable(), thinking that if it was set before then it would have already scheduled unless it was already in the scheduler. This worked fine except in the case of SMP, where another task would set the NEED_RESCHED flag for a task on another CPU, and then kick off an IPI to trigger it. This could cause the NEED_RESCHED to be saved at ftrace_preempt_disable() but the IPI to arrive in the the preempt disabled section. The ftrace_preempt_enable() would not call the scheduler because the flag was already set before entring the section. This bug would cause a missed preemption check and cause lower latencies. Investigating further, I found that the recusion caused by the function tracer was not due to schedule(), but due to preempt_schedule(). Now that preempt_schedule is completely annotated with notrace, the recusion no longer is an issue. Reported-by: Thomas Gleixner Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 55e4851..3572714 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1404,7 +1404,6 @@ int trace_vbprintk(unsigned long ip, const char *fmt, va_list args) struct bprint_entry *entry; unsigned long flags; int disable; - int resched; int cpu, len = 0, size, pc; if (unlikely(tracing_selftest_running || tracing_disabled)) @@ -1414,7 +1413,7 @@ int trace_vbprintk(unsigned long ip, const char *fmt, va_list args) pause_graph_tracing(); pc = preempt_count(); - resched = ftrace_preempt_disable(); + preempt_disable_notrace(); cpu = raw_smp_processor_id(); data = tr->data[cpu]; @@ -1452,7 +1451,7 @@ out_unlock: out: atomic_dec_return(&data->disabled); - ftrace_preempt_enable(resched); + preempt_enable_notrace(); unpause_graph_tracing(); return len; -- cgit v1.1 From 30dbb20e68e6f7df974b77d2350ebad5eb6f6c9e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Am=C3=A9rico=20Wang?= Date: Wed, 26 May 2010 18:57:53 +0800 Subject: tracing: Remove boot tracer The boot tracer is useless. It simply logs the initcalls but in fact these initcalls are also logged through printk while using the initcall_debug kernel parameter. Nobody seem to be using it so far. Then just remove it. Signed-off-by: WANG Cong Cc: Chase Douglas Cc: Steven Rostedt Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Li Zefan LKML-Reference: <20100526105753.GA5677@cr0.nay.redhat.com> [ remove the hooks in main.c, and the headers ] Signed-off-by: Frederic Weisbecker --- kernel/trace/trace.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 55e4851..036fbc2 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -4603,9 +4603,6 @@ __init static int tracer_alloc_buffers(void) register_tracer(&nop_trace); current_trace = &nop_trace; -#ifdef CONFIG_BOOT_TRACER - register_tracer(&boot_tracer); -#endif /* All seems OK, enable tracing */ tracing_disabled = 0; -- cgit v1.1 From f376bf5ffbad863d4bc3b2586b7e34cdf756ad17 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 16 Jul 2010 00:26:26 +0200 Subject: tracing: Remove sysprof ftrace plugin The sysprof ftrace plugin doesn't seem to be seriously used somewhere. There is a branch in the sysprof tree that makes an interface to it, but the real sysprof tool uses either its own module or perf events. Drop the sysprof ftrace plugin then, as it's mostly useless. Signed-off-by: Frederic Weisbecker Acked-by: Soeren Sandmann Cc: Peter Zijlstra Cc: Ingo Molnar Cc: Steven Rostedt Cc: Li Zefan --- kernel/trace/trace.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 8683dec..78a49e6 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -4354,9 +4354,6 @@ static __init int tracer_init_debugfs(void) trace_create_file("dyn_ftrace_total_info", 0444, d_tracer, &ftrace_update_tot_cnt, &tracing_dyn_info_fops); #endif -#ifdef CONFIG_SYSPROF_TRACER - init_tracer_sysprof_debugfs(d_tracer); -#endif create_trace_options_dir(); -- cgit v1.1 From eb7beb5c09af75494234ea6acd09d0a647cf7338 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Fri, 16 Jul 2010 00:50:03 +0200 Subject: tracing: Remove special traces Special traces type was only used by sysprof. Lets remove it now that sysprof ftrace plugin has been dropped. Signed-off-by: Frederic Weisbecker Acked-by: Soeren Sandmann Cc: Peter Zijlstra Cc: Ingo Molnar Cc: Steven Rostedt Cc: Li Zefan --- kernel/trace/trace.c | 55 ---------------------------------------------------- 1 file changed, 55 deletions(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 78a49e6..d9a4aa0 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1331,61 +1331,6 @@ static void __trace_userstack(struct trace_array *tr, unsigned long flags) #endif /* CONFIG_STACKTRACE */ -static void -ftrace_trace_special(void *__tr, - unsigned long arg1, unsigned long arg2, unsigned long arg3, - int pc) -{ - struct ftrace_event_call *call = &event_special; - struct ring_buffer_event *event; - struct trace_array *tr = __tr; - struct ring_buffer *buffer = tr->buffer; - struct special_entry *entry; - - event = trace_buffer_lock_reserve(buffer, TRACE_SPECIAL, - sizeof(*entry), 0, pc); - if (!event) - return; - entry = ring_buffer_event_data(event); - entry->arg1 = arg1; - entry->arg2 = arg2; - entry->arg3 = arg3; - - if (!filter_check_discard(call, entry, buffer, event)) - trace_buffer_unlock_commit(buffer, event, 0, pc); -} - -void -__trace_special(void *__tr, void *__data, - unsigned long arg1, unsigned long arg2, unsigned long arg3) -{ - ftrace_trace_special(__tr, arg1, arg2, arg3, preempt_count()); -} - -void -ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) -{ - struct trace_array *tr = &global_trace; - struct trace_array_cpu *data; - unsigned long flags; - int cpu; - int pc; - - if (tracing_disabled) - return; - - pc = preempt_count(); - local_irq_save(flags); - cpu = raw_smp_processor_id(); - data = tr->data[cpu]; - - if (likely(atomic_inc_return(&data->disabled) == 1)) - ftrace_trace_special(tr, arg1, arg2, arg3, pc); - - atomic_dec(&data->disabled); - local_irq_restore(flags); -} - /** * trace_vbprintk - write binary msg to tracing buffer * -- cgit v1.1 From b444786f1a797a7f84e2561346a670649f9c7b3c Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Wed, 7 Jul 2010 23:40:11 +0200 Subject: tracing: Use generic_file_llseek for debugfs The default for llseek will change to no_llseek, so the tracing debugfs files need to add explicit .llseek assignments. Since we're dealing with regular files from a VFS perspective, use generic_file_llseek. Signed-off-by: Arnd Bergmann Cc: Steven Rostedt Cc: Ingo Molnar Cc: John Kacur Cc: Li Zefan LKML-Reference: <1278538820-1392-10-git-send-email-arnd@arndb.de> Signed-off-by: Frederic Weisbecker --- kernel/trace/trace.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d9a4aa0..c1752da 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2338,6 +2338,7 @@ static const struct file_operations show_traces_fops = { .open = show_traces_open, .read = seq_read, .release = seq_release, + .llseek = seq_lseek, }; /* @@ -2431,6 +2432,7 @@ static const struct file_operations tracing_cpumask_fops = { .open = tracing_open_generic, .read = tracing_cpumask_read, .write = tracing_cpumask_write, + .llseek = generic_file_llseek, }; static int tracing_trace_options_show(struct seq_file *m, void *v) @@ -2597,6 +2599,7 @@ tracing_readme_read(struct file *filp, char __user *ubuf, static const struct file_operations tracing_readme_fops = { .open = tracing_open_generic, .read = tracing_readme_read, + .llseek = generic_file_llseek, }; static ssize_t @@ -2647,6 +2650,7 @@ tracing_saved_cmdlines_read(struct file *file, char __user *ubuf, static const struct file_operations tracing_saved_cmdlines_fops = { .open = tracing_open_generic, .read = tracing_saved_cmdlines_read, + .llseek = generic_file_llseek, }; static ssize_t @@ -2976,6 +2980,7 @@ static int tracing_open_pipe(struct inode *inode, struct file *filp) if (iter->trace->pipe_open) iter->trace->pipe_open(iter); + nonseekable_open(inode, filp); out: mutex_unlock(&trace_types_lock); return ret; @@ -3534,18 +3539,21 @@ static const struct file_operations tracing_max_lat_fops = { .open = tracing_open_generic, .read = tracing_max_lat_read, .write = tracing_max_lat_write, + .llseek = generic_file_llseek, }; static const struct file_operations tracing_ctrl_fops = { .open = tracing_open_generic, .read = tracing_ctrl_read, .write = tracing_ctrl_write, + .llseek = generic_file_llseek, }; static const struct file_operations set_tracer_fops = { .open = tracing_open_generic, .read = tracing_set_trace_read, .write = tracing_set_trace_write, + .llseek = generic_file_llseek, }; static const struct file_operations tracing_pipe_fops = { @@ -3554,17 +3562,20 @@ static const struct file_operations tracing_pipe_fops = { .read = tracing_read_pipe, .splice_read = tracing_splice_read_pipe, .release = tracing_release_pipe, + .llseek = no_llseek, }; static const struct file_operations tracing_entries_fops = { .open = tracing_open_generic, .read = tracing_entries_read, .write = tracing_entries_write, + .llseek = generic_file_llseek, }; static const struct file_operations tracing_mark_fops = { .open = tracing_open_generic, .write = tracing_mark_write, + .llseek = generic_file_llseek, }; static const struct file_operations trace_clock_fops = { @@ -3870,6 +3881,7 @@ tracing_stats_read(struct file *filp, char __user *ubuf, static const struct file_operations tracing_stats_fops = { .open = tracing_open_generic, .read = tracing_stats_read, + .llseek = generic_file_llseek, }; #ifdef CONFIG_DYNAMIC_FTRACE @@ -3906,6 +3918,7 @@ tracing_read_dyn_info(struct file *filp, char __user *ubuf, static const struct file_operations tracing_dyn_info_fops = { .open = tracing_open_generic, .read = tracing_read_dyn_info, + .llseek = generic_file_llseek, }; #endif @@ -4059,6 +4072,7 @@ static const struct file_operations trace_options_fops = { .open = tracing_open_generic, .read = trace_options_read, .write = trace_options_write, + .llseek = generic_file_llseek, }; static ssize_t @@ -4110,6 +4124,7 @@ static const struct file_operations trace_options_core_fops = { .open = tracing_open_generic, .read = trace_options_core_read, .write = trace_options_core_write, + .llseek = generic_file_llseek, }; struct dentry *trace_create_file(const char *name, -- cgit v1.1 From e870e9a1240bcef1157ffaaf71dac63362e71904 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Fri, 2 Jul 2010 11:07:32 +0800 Subject: tracing: Allow to disable cmdline recording We found that even enabling a single trace event that will rarely be triggered can add big overhead to context switch. (lmbench context switch test) ------------------------------------------------- 2p/0K 2p/16K 2p/64K 8p/16K 8p/64K 16p/16K 16p/64K ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw ctxsw ------ ------ ------ ------ ------ ------- ------- 2.19 2.3 2.21 2.56 2.13 2.54 2.07 2.39 2.51 2.35 2.75 2.27 2.81 2.24 The overhead is 6% ~ 11%. It's because when a trace event is enabled 3 tracepoints (sched_switch, sched_wakeup, sched_wakeup_new) will be activated to map pid to cmdname. We'd like to avoid this overhead, so add a trace option '(no)record-cmd' to allow to disable cmdline recording. Signed-off-by: Li Zefan LKML-Reference: <4C2D57F4.2050204@cn.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 8683dec..af90429 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -344,7 +344,7 @@ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); /* trace_flags holds trace_options default values */ unsigned long trace_flags = TRACE_ITER_PRINT_PARENT | TRACE_ITER_PRINTK | TRACE_ITER_ANNOTATE | TRACE_ITER_CONTEXT_INFO | TRACE_ITER_SLEEP_TIME | - TRACE_ITER_GRAPH_TIME; + TRACE_ITER_GRAPH_TIME | TRACE_ITER_RECORD_CMD; static int trace_stop_count; static DEFINE_SPINLOCK(tracing_start_lock); @@ -428,6 +428,7 @@ static const char *trace_options[] = { "latency-format", "sleep-time", "graph-time", + "record-cmd", NULL }; @@ -2561,6 +2562,9 @@ static void set_tracer_flags(unsigned int mask, int enabled) trace_flags |= mask; else trace_flags &= ~mask; + + if (mask == TRACE_ITER_RECORD_CMD) + trace_event_enable_cmd_record(enabled); } static ssize_t -- cgit v1.1 From ef710e100c1068d3dd5774d2b34c5485219e06ce Mon Sep 17 00:00:00 2001 From: KOSAKI Motohiro Date: Thu, 1 Jul 2010 14:34:35 +0900 Subject: tracing: Shrink max latency ringbuffer if unnecessary Documentation/trace/ftrace.txt says buffer_size_kb: This sets or displays the number of kilobytes each CPU buffer can hold. The tracer buffers are the same size for each CPU. The displayed number is the size of the CPU buffer and not total size of all buffers. The trace buffers are allocated in pages (blocks of memory that the kernel uses for allocation, usually 4 KB in size). If the last page allocated has room for more bytes than requested, the rest of the page will be used, making the actual allocation bigger than requested. ( Note, the size may not be a multiple of the page size due to buffer management overhead. ) This can only be updated when the current_tracer is set to "nop". But it's incorrect. currently total memory consumption is 'buffer_size_kb x CPUs x 2'. Why two times difference is there? because ftrace implicitly allocate the buffer for max latency too. That makes sad result when admin want to use large buffer. (If admin want full logging and makes detail analysis). example, If admin have 24 CPUs machine and write 200MB to buffer_size_kb, the system consume ~10GB memory (200MB x 24 x 2). umm.. 5GB memory waste is usually unacceptable. Fortunatelly, almost all users don't use max latency feature. The max latency buffer can be disabled easily. This patch shrink buffer size of the max latency buffer if unnecessary. Signed-off-by: KOSAKI Motohiro LKML-Reference: <20100701104554.DA2D.A69D9226@jp.fujitsu.com> Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 38 ++++++++++++++++++++++++++++++++------ 1 file changed, 32 insertions(+), 6 deletions(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index af90429..f7488f4 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -660,6 +660,10 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) return; WARN_ON_ONCE(!irqs_disabled()); + if (!current_trace->use_max_tr) { + WARN_ON_ONCE(1); + return; + } arch_spin_lock(&ftrace_max_lock); tr->buffer = max_tr.buffer; @@ -686,6 +690,11 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) return; WARN_ON_ONCE(!irqs_disabled()); + if (!current_trace->use_max_tr) { + WARN_ON_ONCE(1); + return; + } + arch_spin_lock(&ftrace_max_lock); ftrace_disable_cpu(); @@ -2801,6 +2810,9 @@ static int tracing_resize_ring_buffer(unsigned long size) if (ret < 0) return ret; + if (!current_trace->use_max_tr) + goto out; + ret = ring_buffer_resize(max_tr.buffer, size); if (ret < 0) { int r; @@ -2828,11 +2840,14 @@ static int tracing_resize_ring_buffer(unsigned long size) return ret; } + max_tr.entries = size; + out: global_trace.entries = size; return ret; } + /** * tracing_update_buffers - used by tracing facility to expand ring buffers * @@ -2893,12 +2908,26 @@ static int tracing_set_tracer(const char *buf) trace_branch_disable(); if (current_trace && current_trace->reset) current_trace->reset(tr); - + if (current_trace && current_trace->use_max_tr) { + /* + * We don't free the ring buffer. instead, resize it because + * The max_tr ring buffer has some state (e.g. ring->clock) and + * we want preserve it. + */ + ring_buffer_resize(max_tr.buffer, 1); + max_tr.entries = 1; + } destroy_trace_option_files(topts); current_trace = t; topts = create_trace_option_files(current_trace); + if (current_trace->use_max_tr) { + ret = ring_buffer_resize(max_tr.buffer, global_trace.entries); + if (ret < 0) + goto out; + max_tr.entries = global_trace.entries; + } if (t->init) { ret = tracer_init(t, tr); @@ -3480,7 +3509,6 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, } tracing_start(); - max_tr.entries = global_trace.entries; mutex_unlock(&trace_types_lock); return cnt; @@ -4578,16 +4606,14 @@ __init static int tracer_alloc_buffers(void) #ifdef CONFIG_TRACER_MAX_TRACE - max_tr.buffer = ring_buffer_alloc(ring_buf_size, - TRACE_BUFFER_FLAGS); + max_tr.buffer = ring_buffer_alloc(1, TRACE_BUFFER_FLAGS); if (!max_tr.buffer) { printk(KERN_ERR "tracer: failed to allocate max ring buffer!\n"); WARN_ON(1); ring_buffer_free(global_trace.buffer); goto out_free_cpumask; } - max_tr.entries = ring_buffer_size(max_tr.buffer); - WARN_ON(max_tr.entries != global_trace.entries); + max_tr.entries = 1; #endif /* Allocate the first page for all buffers */ -- cgit v1.1 From 24a461d537f49f9da6533d83100999ea08c6c755 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Sat, 10 Jul 2010 12:06:44 +0200 Subject: trace: strlen() return doesn't account for the NULL We need to add one to the strlen() return because of the NULL character. The type->name here generally comes from the kernel and I don't think any of them come close to being MAX_TRACER_SIZE (100) characters long so this is basically a cleanup. Signed-off-by: Dan Carpenter LKML-Reference: <20100710100644.GV19184@bicker> Signed-off-by: Steven Rostedt --- kernel/trace/trace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel/trace/trace.c') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index f7488f4..cacb6f0 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -739,7 +739,7 @@ __acquires(kernel_lock) return -1; } - if (strlen(type->name) > MAX_TRACER_SIZE) { + if (strlen(type->name) >= MAX_TRACER_SIZE) { pr_info("Tracer has a name longer than %d\n", MAX_TRACER_SIZE); return -1; } -- cgit v1.1