From bc078e4eab65f11bbaeed380593ab8151b30d703 Mon Sep 17 00:00:00 2001 From: Martin Schwidefsky Date: Tue, 2 Mar 2010 16:01:10 +0100 Subject: oprofile: convert oprofile from timer_hook to hrtimer Oprofile is currently broken on systems running with NOHZ enabled. A maximum of 1 tick is accounted via the timer_hook if a cpu sleeps for a longer period of time. This does bad things to the percentages in the profiler output. To solve this problem convert oprofile to use a restarting hrtimer instead of the timer_hook. Signed-off-by: Martin Schwidefsky Signed-off-by: Robert Richter --- drivers/oprofile/oprof.c | 12 ++++--- drivers/oprofile/oprof.h | 3 +- drivers/oprofile/timer_int.c | 78 +++++++++++++++++++++++++++++++++++++++----- 3 files changed, 79 insertions(+), 14 deletions(-) (limited to 'drivers') diff --git a/drivers/oprofile/oprof.c b/drivers/oprofile/oprof.c index dc8a0428260d..b336cd9ee7a1 100644 --- a/drivers/oprofile/oprof.c +++ b/drivers/oprofile/oprof.c @@ -253,22 +253,26 @@ static int __init oprofile_init(void) int err; err = oprofile_arch_init(&oprofile_ops); - if (err < 0 || timer) { printk(KERN_INFO "oprofile: using timer interrupt.\n"); - oprofile_timer_init(&oprofile_ops); + err = oprofile_timer_init(&oprofile_ops); + if (err) + goto out_arch; } - err = oprofilefs_register(); if (err) - oprofile_arch_exit(); + goto out_arch; + return 0; +out_arch: + oprofile_arch_exit(); return err; } static void __exit oprofile_exit(void) { + oprofile_timer_exit(); oprofilefs_unregister(); oprofile_arch_exit(); } diff --git a/drivers/oprofile/oprof.h b/drivers/oprofile/oprof.h index cb92f5c98c1a..47e12cb4ee8b 100644 --- a/drivers/oprofile/oprof.h +++ b/drivers/oprofile/oprof.h @@ -34,7 +34,8 @@ struct super_block; struct dentry; void oprofile_create_files(struct super_block *sb, struct dentry *root); -void oprofile_timer_init(struct oprofile_operations *ops); +int oprofile_timer_init(struct oprofile_operations *ops); +void oprofile_timer_exit(void); int oprofile_set_backtrace(unsigned long depth); int oprofile_set_timeout(unsigned long time); diff --git a/drivers/oprofile/timer_int.c b/drivers/oprofile/timer_int.c index 333f915568c7..dc0ae4d14dff 100644 --- a/drivers/oprofile/timer_int.c +++ b/drivers/oprofile/timer_int.c @@ -13,34 +13,94 @@ #include #include #include +#include +#include +#include #include #include "oprof.h" -static int timer_notify(struct pt_regs *regs) +static DEFINE_PER_CPU(struct hrtimer, oprofile_hrtimer); + +static enum hrtimer_restart oprofile_hrtimer_notify(struct hrtimer *hrtimer) +{ + oprofile_add_sample(get_irq_regs(), 0); + hrtimer_forward_now(hrtimer, ns_to_ktime(TICK_NSEC)); + return HRTIMER_RESTART; +} + +static void __oprofile_hrtimer_start(void *unused) +{ + struct hrtimer *hrtimer = &__get_cpu_var(oprofile_hrtimer); + + hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hrtimer->function = oprofile_hrtimer_notify; + + hrtimer_start(hrtimer, ns_to_ktime(TICK_NSEC), + HRTIMER_MODE_REL_PINNED); +} + +static int oprofile_hrtimer_start(void) { - oprofile_add_sample(regs, 0); + on_each_cpu(__oprofile_hrtimer_start, NULL, 1); return 0; } -static int timer_start(void) +static void __oprofile_hrtimer_stop(int cpu) { - return register_timer_hook(timer_notify); + struct hrtimer *hrtimer = &per_cpu(oprofile_hrtimer, cpu); + + hrtimer_cancel(hrtimer); } +static void oprofile_hrtimer_stop(void) +{ + int cpu; + + for_each_online_cpu(cpu) + __oprofile_hrtimer_stop(cpu); +} -static void timer_stop(void) +static int __cpuinit oprofile_cpu_notify(struct notifier_block *self, + unsigned long action, void *hcpu) { - unregister_timer_hook(timer_notify); + long cpu = (long) hcpu; + + switch (action) { + case CPU_ONLINE: + case CPU_ONLINE_FROZEN: + smp_call_function_single(cpu, __oprofile_hrtimer_start, + NULL, 1); + break; + case CPU_DEAD: + case CPU_DEAD_FROZEN: + __oprofile_hrtimer_stop(cpu); + break; + } + return NOTIFY_OK; } +static struct notifier_block __refdata oprofile_cpu_notifier = { + .notifier_call = oprofile_cpu_notify, +}; -void __init oprofile_timer_init(struct oprofile_operations *ops) +int __init oprofile_timer_init(struct oprofile_operations *ops) { + int rc; + + rc = register_hotcpu_notifier(&oprofile_cpu_notifier); + if (rc) + return rc; ops->create_files = NULL; ops->setup = NULL; ops->shutdown = NULL; - ops->start = timer_start; - ops->stop = timer_stop; + ops->start = oprofile_hrtimer_start; + ops->stop = oprofile_hrtimer_stop; ops->cpu_type = "timer"; + return 0; +} + +void __exit oprofile_timer_exit(void) +{ + unregister_hotcpu_notifier(&oprofile_cpu_notifier); } -- cgit v1.2.3 From 66a8cb95ed04025664d1db4e952155ee1dccd048 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 31 Mar 2010 13:21:56 -0400 Subject: ring-buffer: Add place holder recording of dropped events Currently, when the ring buffer drops events, it does not record the fact that it did so. It does inform the writer that the event was dropped by returning a NULL event, but it does not put in any place holder where the event was dropped. This is not a trivial thing to add because the ring buffer mostly runs in overwrite (flight recorder) mode. That is, when the ring buffer is full, new data will overwrite old data. In a produce/consumer mode, where new data is simply dropped when the ring buffer is full, it is trivial to add the placeholder for dropped events. When there's more room to write new data, then a special event can be added to notify the reader about the dropped events. But in overwrite mode, any new write can overwrite events. A place holder can not be inserted into the ring buffer since there never may be room. A reader could also come in at anytime and miss the placeholder. Luckily, the way the ring buffer works, the read side can find out if events were lost or not, and how many events. Everytime a write takes place, if it overwrites the header page (the next read) it updates a "overrun" variable that keeps track of the number of lost events. When a reader swaps out a page from the ring buffer, it can record this number, perfom the swap, and then check to see if the number changed, and take the diff if it has, which would be the number of events dropped. This can be stored by the reader and returned to callers of the reader. Since the reader page swap will fail if the writer moved the head page since the time the reader page set up the swap, this gives room to record the overruns without worrying about races. If the reader sets up the pages, records the overrun, than performs the swap, if the swap succeeds, then the overrun variable has not been updated since the setup before the swap. For binary readers of the ring buffer, a flag is set in the header of each sub page (sub buffer) of the ring buffer. This flag is embedded in the size field of the data on the sub buffer, in the 31st bit (the size can be 32 or 64 bits depending on the architecture), but only 27 bits needs to be used for the actual size (less actually). We could add a new field in the sub buffer header to also record the number of events dropped since the last read, but this will change the format of the binary ring buffer a bit too much. Perhaps this change can be made if the information on the number of events dropped is considered important enough. Note, the notification of dropped events is only used by consuming reads or peeking at the ring buffer. Iterating over the ring buffer does not keep this information because the necessary data is only available when a page swap is made, and the iterator does not swap out pages. Cc: Robert Richter Cc: Andi Kleen Cc: Li Zefan Cc: Arnaldo Carvalho de Melo Cc: "Luis Claudio R. Goncalves" Cc: Frederic Weisbecker Signed-off-by: Steven Rostedt --- drivers/oprofile/cpu_buffer.c | 4 +- include/linux/ring_buffer.h | 6 ++- kernel/trace/ring_buffer.c | 72 +++++++++++++++++++++++++++++++++--- kernel/trace/ring_buffer_benchmark.c | 2 +- kernel/trace/trace.c | 4 +- kernel/trace/trace_functions_graph.c | 5 ++- kernel/trace/trace_selftest.c | 2 +- 7 files changed, 79 insertions(+), 16 deletions(-) (limited to 'drivers') diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c index 166b67ea622f..7581dbe456da 100644 --- a/drivers/oprofile/cpu_buffer.c +++ b/drivers/oprofile/cpu_buffer.c @@ -186,14 +186,14 @@ int op_cpu_buffer_write_commit(struct op_entry *entry) struct op_sample *op_cpu_buffer_read_entry(struct op_entry *entry, int cpu) { struct ring_buffer_event *e; - e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL); + e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL, NULL); if (e) goto event; if (ring_buffer_swap_cpu(op_ring_buffer_read, op_ring_buffer_write, cpu)) return NULL; - e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL); + e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL, NULL); if (e) goto event; return NULL; diff --git a/include/linux/ring_buffer.h b/include/linux/ring_buffer.h index 5fcc31ed5771..c8297761e414 100644 --- a/include/linux/ring_buffer.h +++ b/include/linux/ring_buffer.h @@ -120,9 +120,11 @@ int ring_buffer_write(struct ring_buffer *buffer, unsigned long length, void *data); struct ring_buffer_event * -ring_buffer_peek(struct ring_buffer *buffer, int cpu, u64 *ts); +ring_buffer_peek(struct ring_buffer *buffer, int cpu, u64 *ts, + unsigned long *lost_events); struct ring_buffer_event * -ring_buffer_consume(struct ring_buffer *buffer, int cpu, u64 *ts); +ring_buffer_consume(struct ring_buffer *buffer, int cpu, u64 *ts, + unsigned long *lost_events); struct ring_buffer_iter * ring_buffer_read_start(struct ring_buffer *buffer, int cpu); diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index d1187ef20caf..8295650444c5 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -318,6 +318,9 @@ EXPORT_SYMBOL_GPL(ring_buffer_event_data); #define TS_MASK ((1ULL << TS_SHIFT) - 1) #define TS_DELTA_TEST (~TS_MASK) +/* Flag when events were overwritten */ +#define RB_MISSED_EVENTS (1 << 31) + struct buffer_data_page { u64 time_stamp; /* page time stamp */ local_t commit; /* write committed index */ @@ -416,6 +419,12 @@ int ring_buffer_print_page_header(struct trace_seq *s) (unsigned int)sizeof(field.commit), (unsigned int)is_signed_type(long)); + ret = trace_seq_printf(s, "\tfield: int overwrite;\t" + "offset:%u;\tsize:%u;\tsigned:%u;\n", + (unsigned int)offsetof(typeof(field), commit), + 1, + (unsigned int)is_signed_type(long)); + ret = trace_seq_printf(s, "\tfield: char data;\t" "offset:%u;\tsize:%u;\tsigned:%u;\n", (unsigned int)offsetof(typeof(field), data), @@ -439,6 +448,8 @@ struct ring_buffer_per_cpu { struct buffer_page *tail_page; /* write to tail */ struct buffer_page *commit_page; /* committed pages */ struct buffer_page *reader_page; + unsigned long lost_events; + unsigned long last_overrun; local_t commit_overrun; local_t overrun; local_t entries; @@ -2835,6 +2846,7 @@ static struct buffer_page * rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer) { struct buffer_page *reader = NULL; + unsigned long overwrite; unsigned long flags; int nr_loops = 0; int ret; @@ -2895,6 +2907,18 @@ rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer) /* The reader page will be pointing to the new head */ rb_set_list_to_head(cpu_buffer, &cpu_buffer->reader_page->list); + /* + * We want to make sure we read the overruns after we set up our + * pointers to the next object. The writer side does a + * cmpxchg to cross pages which acts as the mb on the writer + * side. Note, the reader will constantly fail the swap + * while the writer is updating the pointers, so this + * guarantees that the overwrite recorded here is the one we + * want to compare with the last_overrun. + */ + smp_mb(); + overwrite = local_read(&(cpu_buffer->overrun)); + /* * Here's the tricky part. * @@ -2926,6 +2950,11 @@ rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer) cpu_buffer->reader_page = reader; rb_reset_reader_page(cpu_buffer); + if (overwrite != cpu_buffer->last_overrun) { + cpu_buffer->lost_events = overwrite - cpu_buffer->last_overrun; + cpu_buffer->last_overrun = overwrite; + } + goto again; out: @@ -3002,8 +3031,14 @@ static void rb_advance_iter(struct ring_buffer_iter *iter) rb_advance_iter(iter); } +static int rb_lost_events(struct ring_buffer_per_cpu *cpu_buffer) +{ + return cpu_buffer->lost_events; +} + static struct ring_buffer_event * -rb_buffer_peek(struct ring_buffer_per_cpu *cpu_buffer, u64 *ts) +rb_buffer_peek(struct ring_buffer_per_cpu *cpu_buffer, u64 *ts, + unsigned long *lost_events) { struct ring_buffer_event *event; struct buffer_page *reader; @@ -3055,6 +3090,8 @@ rb_buffer_peek(struct ring_buffer_per_cpu *cpu_buffer, u64 *ts) ring_buffer_normalize_time_stamp(cpu_buffer->buffer, cpu_buffer->cpu, ts); } + if (lost_events) + *lost_events = rb_lost_events(cpu_buffer); return event; default: @@ -3165,12 +3202,14 @@ static inline int rb_ok_to_lock(void) * @buffer: The ring buffer to read * @cpu: The cpu to peak at * @ts: The timestamp counter of this event. + * @lost_events: a variable to store if events were lost (may be NULL) * * This will return the event that will be read next, but does * not consume the data. */ struct ring_buffer_event * -ring_buffer_peek(struct ring_buffer *buffer, int cpu, u64 *ts) +ring_buffer_peek(struct ring_buffer *buffer, int cpu, u64 *ts, + unsigned long *lost_events) { struct ring_buffer_per_cpu *cpu_buffer = buffer->buffers[cpu]; struct ring_buffer_event *event; @@ -3185,7 +3224,7 @@ ring_buffer_peek(struct ring_buffer *buffer, int cpu, u64 *ts) local_irq_save(flags); if (dolock) spin_lock(&cpu_buffer->reader_lock); - event = rb_buffer_peek(cpu_buffer, ts); + event = rb_buffer_peek(cpu_buffer, ts, lost_events); if (event && event->type_len == RINGBUF_TYPE_PADDING) rb_advance_reader(cpu_buffer); if (dolock) @@ -3227,13 +3266,17 @@ ring_buffer_iter_peek(struct ring_buffer_iter *iter, u64 *ts) /** * ring_buffer_consume - return an event and consume it * @buffer: The ring buffer to get the next event from + * @cpu: the cpu to read the buffer from + * @ts: a variable to store the timestamp (may be NULL) + * @lost_events: a variable to store if events were lost (may be NULL) * * Returns the next event in the ring buffer, and that event is consumed. * Meaning, that sequential reads will keep returning a different event, * and eventually empty the ring buffer if the producer is slower. */ struct ring_buffer_event * -ring_buffer_consume(struct ring_buffer *buffer, int cpu, u64 *ts) +ring_buffer_consume(struct ring_buffer *buffer, int cpu, u64 *ts, + unsigned long *lost_events) { struct ring_buffer_per_cpu *cpu_buffer; struct ring_buffer_event *event = NULL; @@ -3254,9 +3297,11 @@ ring_buffer_consume(struct ring_buffer *buffer, int cpu, u64 *ts) if (dolock) spin_lock(&cpu_buffer->reader_lock); - event = rb_buffer_peek(cpu_buffer, ts); - if (event) + event = rb_buffer_peek(cpu_buffer, ts, lost_events); + if (event) { + cpu_buffer->lost_events = 0; rb_advance_reader(cpu_buffer); + } if (dolock) spin_unlock(&cpu_buffer->reader_lock); @@ -3405,6 +3450,9 @@ rb_reset_cpu(struct ring_buffer_per_cpu *cpu_buffer) cpu_buffer->write_stamp = 0; cpu_buffer->read_stamp = 0; + cpu_buffer->lost_events = 0; + cpu_buffer->last_overrun = 0; + rb_head_page_activate(cpu_buffer); } @@ -3684,6 +3732,7 @@ int ring_buffer_read_page(struct ring_buffer *buffer, unsigned int commit; unsigned int read; u64 save_timestamp; + int missed_events = 0; int ret = -1; if (!cpumask_test_cpu(cpu, buffer->cpumask)) @@ -3716,6 +3765,10 @@ int ring_buffer_read_page(struct ring_buffer *buffer, read = reader->read; commit = rb_page_commit(reader); + /* Check if any events were dropped */ + if (cpu_buffer->lost_events) + missed_events = 1; + /* * If this page has been partially read or * if len is not big enough to read the rest of the page or @@ -3779,6 +3832,13 @@ int ring_buffer_read_page(struct ring_buffer *buffer, } ret = read; + cpu_buffer->lost_events = 0; + /* + * Set a flag in the commit field if we lost events + */ + if (missed_events) + local_add(RB_MISSED_EVENTS, &bpage->commit); + out_unlock: spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags); diff --git a/kernel/trace/ring_buffer_benchmark.c b/kernel/trace/ring_buffer_benchmark.c index df74c7982255..dc56556b55a2 100644 --- a/kernel/trace/ring_buffer_benchmark.c +++ b/kernel/trace/ring_buffer_benchmark.c @@ -81,7 +81,7 @@ static enum event_status read_event(int cpu) int *entry; u64 ts; - event = ring_buffer_consume(buffer, cpu, &ts); + event = ring_buffer_consume(buffer, cpu, &ts, NULL); if (!event) return EVENT_DROPPED; diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3ec2ee6f6560..fabb0033a9be 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1556,7 +1556,7 @@ peek_next_entry(struct trace_iterator *iter, int cpu, u64 *ts) if (buf_iter) event = ring_buffer_iter_peek(buf_iter, ts); else - event = ring_buffer_peek(iter->tr->buffer, cpu, ts); + event = ring_buffer_peek(iter->tr->buffer, cpu, ts, NULL); ftrace_enable_cpu(); @@ -1635,7 +1635,7 @@ static void trace_consume(struct trace_iterator *iter) { /* Don't allow ftrace to trace into the ring buffers */ ftrace_disable_cpu(); - ring_buffer_consume(iter->tr->buffer, iter->cpu, &iter->ts); + ring_buffer_consume(iter->tr->buffer, iter->cpu, &iter->ts, NULL); ftrace_enable_cpu(); } diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index e6989d9b44da..a7f75fb10aa4 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -489,9 +489,10 @@ get_return_for_leaf(struct trace_iterator *iter, * We need to consume the current entry to see * the next one. */ - ring_buffer_consume(iter->tr->buffer, iter->cpu, NULL); + ring_buffer_consume(iter->tr->buffer, iter->cpu, + NULL, NULL); event = ring_buffer_peek(iter->tr->buffer, iter->cpu, - NULL); + NULL, NULL); } if (!event) diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 280fea470d67..e50180874c63 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -29,7 +29,7 @@ static int trace_test_buffer_cpu(struct trace_array *tr, int cpu) struct trace_entry *entry; unsigned int loops = 0; - while ((event = ring_buffer_consume(tr->buffer, cpu, NULL))) { + while ((event = ring_buffer_consume(tr->buffer, cpu, NULL, NULL))) { entry = ring_buffer_event_data(event); /* -- cgit v1.2.3 From cb6e943ccf19ab6d3189147e9d625a992e016084 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Thu, 1 Apr 2010 03:17:25 +0200 Subject: oprofile: remove double ring buffering oprofile used a double buffer scheme for its cpu event buffer to avoid races on reading with the old locked ring buffer. But that is obsolete now with the new ring buffer, so simply use a single buffer. This greatly simplifies the code and avoids a lot of sample drops on large runs, especially with call graph. Based on suggestions from Steven Rostedt For stable kernels from v2.6.32, but not earlier. Signed-off-by: Andi Kleen Cc: Steven Rostedt Cc: stable Signed-off-by: Robert Richter --- drivers/oprofile/cpu_buffer.c | 63 +++++++++---------------------------------- 1 file changed, 13 insertions(+), 50 deletions(-) (limited to 'drivers') diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c index 166b67ea622f..de82183bb9b3 100644 --- a/drivers/oprofile/cpu_buffer.c +++ b/drivers/oprofile/cpu_buffer.c @@ -30,23 +30,7 @@ #define OP_BUFFER_FLAGS 0 -/* - * Read and write access is using spin locking. Thus, writing to the - * buffer by NMI handler (x86) could occur also during critical - * sections when reading the buffer. To avoid this, there are 2 - * buffers for independent read and write access. Read access is in - * process context only, write access only in the NMI handler. If the - * read buffer runs empty, both buffers are swapped atomically. There - * is potentially a small window during swapping where the buffers are - * disabled and samples could be lost. - * - * Using 2 buffers is a little bit overhead, but the solution is clear - * and does not require changes in the ring buffer implementation. It - * can be changed to a single buffer solution when the ring buffer - * access is implemented as non-locking atomic code. - */ -static struct ring_buffer *op_ring_buffer_read; -static struct ring_buffer *op_ring_buffer_write; +static struct ring_buffer *op_ring_buffer; DEFINE_PER_CPU(struct oprofile_cpu_buffer, op_cpu_buffer); static void wq_sync_buffer(struct work_struct *work); @@ -68,12 +52,9 @@ void oprofile_cpu_buffer_inc_smpl_lost(void) void free_cpu_buffers(void) { - if (op_ring_buffer_read) - ring_buffer_free(op_ring_buffer_read); - op_ring_buffer_read = NULL; - if (op_ring_buffer_write) - ring_buffer_free(op_ring_buffer_write); - op_ring_buffer_write = NULL; + if (op_ring_buffer) + ring_buffer_free(op_ring_buffer); + op_ring_buffer = NULL; } #define RB_EVENT_HDR_SIZE 4 @@ -86,11 +67,8 @@ int alloc_cpu_buffers(void) unsigned long byte_size = buffer_size * (sizeof(struct op_sample) + RB_EVENT_HDR_SIZE); - op_ring_buffer_read = ring_buffer_alloc(byte_size, OP_BUFFER_FLAGS); - if (!op_ring_buffer_read) - goto fail; - op_ring_buffer_write = ring_buffer_alloc(byte_size, OP_BUFFER_FLAGS); - if (!op_ring_buffer_write) + op_ring_buffer = ring_buffer_alloc(byte_size, OP_BUFFER_FLAGS); + if (!op_ring_buffer) goto fail; for_each_possible_cpu(i) { @@ -162,16 +140,11 @@ struct op_sample *op_cpu_buffer_write_reserve(struct op_entry *entry, unsigned long size) { entry->event = ring_buffer_lock_reserve - (op_ring_buffer_write, sizeof(struct op_sample) + + (op_ring_buffer, sizeof(struct op_sample) + size * sizeof(entry->sample->data[0])); - if (entry->event) - entry->sample = ring_buffer_event_data(entry->event); - else - entry->sample = NULL; - - if (!entry->sample) + if (!entry->event) return NULL; - + entry->sample = ring_buffer_event_data(entry->event); entry->size = size; entry->data = entry->sample->data; @@ -180,25 +153,16 @@ struct op_sample int op_cpu_buffer_write_commit(struct op_entry *entry) { - return ring_buffer_unlock_commit(op_ring_buffer_write, entry->event); + return ring_buffer_unlock_commit(op_ring_buffer, entry->event); } struct op_sample *op_cpu_buffer_read_entry(struct op_entry *entry, int cpu) { struct ring_buffer_event *e; - e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL); - if (e) - goto event; - if (ring_buffer_swap_cpu(op_ring_buffer_read, - op_ring_buffer_write, - cpu)) + e = ring_buffer_consume(op_ring_buffer, cpu, NULL); + if (!e) return NULL; - e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL); - if (e) - goto event; - return NULL; -event: entry->event = e; entry->sample = ring_buffer_event_data(e); entry->size = (ring_buffer_event_length(e) - sizeof(struct op_sample)) @@ -209,8 +173,7 @@ event: unsigned long op_cpu_buffer_entries(int cpu) { - return ring_buffer_entries_cpu(op_ring_buffer_read, cpu) - + ring_buffer_entries_cpu(op_ring_buffer_write, cpu); + return ring_buffer_entries_cpu(op_ring_buffer, cpu); } static int -- cgit v1.2.3 From 9414e99672271adcc661f3c160a30b374179b92f Mon Sep 17 00:00:00 2001 From: Phil Carmody Date: Wed, 28 Apr 2010 12:09:16 -0500 Subject: oprofile: protect from not being in an IRQ context http://lkml.org/lkml/2010/4/27/285 Protect against dereferencing regs when it's NULL, and force a magic number into pc to prevent too deep processing. This approach permits the dropped samples to be tallied as invalid Instruction Pointer events. e.g. output from about 15mins at 10kHz sample rate: Nr. samples received: 2565380 Nr. samples lost invalid pc: 4 Signed-off-by: Phil Carmody Signed-off-by: Robert Richter --- drivers/oprofile/cpu_buffer.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'drivers') diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c index 0ac8b065ee02..219f79e2210a 100644 --- a/drivers/oprofile/cpu_buffer.c +++ b/drivers/oprofile/cpu_buffer.c @@ -319,8 +319,16 @@ void oprofile_add_ext_sample(unsigned long pc, struct pt_regs * const regs, void oprofile_add_sample(struct pt_regs * const regs, unsigned long event) { - int is_kernel = !user_mode(regs); - unsigned long pc = profile_pc(regs); + int is_kernel; + unsigned long pc; + + if (likely(regs)) { + is_kernel = !user_mode(regs); + pc = profile_pc(regs); + } else { + is_kernel = 0; /* This value will not be used */ + pc = ESCAPE_CODE; /* as this causes an early return. */ + } __oprofile_add_ext_sample(pc, regs, event, is_kernel); } -- cgit v1.2.3