tracing: infrastructure for supporting binary record

Impact: save on memory for tracing

Current tracers are typically using a struct(like struct ftrace_entry,
struct ctx_switch_entry, struct special_entr etc...)to record a binary
event. These structs can only record a their own kind of events.
A new kind of tracer need a new struct and a lot of code too handle it.

So we need a generic binary record for events. This infrastructure
is for this purpose.

[fweisbec@gmail.com: rebase against latest -tip, make it safe while sched
tracing as reported by Steven Rostedt]

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
LKML-Reference: <1236356510-8381-3-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

6 files changed, 237 insertions, 0 deletions

diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 058d949a3214..ad8d3617d0a6 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -97,6 +97,12 @@ config FUNCTION_GRAPH_TRACER
           This is done by setting the current return address on the current
           task structure into a stack of calls.
 
+config TRACE_BPRINTK
+        bool "Binary printk for tracing"
+        default y
+        depends on TRACING
+        select BINARY_PRINTF
+
 config IRQSOFF_TRACER
         bool "Interrupts-off Latency Tracer"
         default n
diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
index f44736c7574a..46557ef4c379 100644
--- a/kernel/trace/Makefile
+++ b/kernel/trace/Makefile
@@ -22,6 +22,7 @@ obj-$(CONFIG_TRACING) += trace.o
 obj-$(CONFIG_TRACING) += trace_clock.o
 obj-$(CONFIG_TRACING) += trace_output.o
 obj-$(CONFIG_TRACING) += trace_stat.o
+obj-$(CONFIG_TRACE_BPRINTK) += trace_bprintk.o
 obj-$(CONFIG_CONTEXT_SWITCH_TRACER) += trace_sched_switch.o
 obj-$(CONFIG_SYSPROF_TRACER) += trace_sysprof.o
 obj-$(CONFIG_FUNCTION_TRACER) += trace_functions.o
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index e6144acf2b75..ff53509e19f8 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -3792,6 +3792,62 @@ int __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap)
 }
 EXPORT_SYMBOL_GPL(__ftrace_vprintk);
 
+/**
+ * trace_vbprintk - write binary msg to tracing buffer
+ *
+ * Caller must insure @fmt are valid when msg is in tracing buffer.
+ */
+int trace_vbprintk(unsigned long ip, const char *fmt, va_list args)
+{
+        static DEFINE_SPINLOCK(trace_buf_lock);
+        static u32 trace_buf[TRACE_BUF_SIZE];
+
+        struct ring_buffer_event *event;
+        struct trace_array *tr = &global_trace;
+        struct trace_array_cpu *data;
+        struct bprintk_entry *entry;
+        unsigned long flags;
+        int resched;
+        int cpu, len = 0, size, pc;
+
+        if (tracing_disabled || !trace_bprintk_enable)
+                return 0;
+
+        pc = preempt_count();
+        resched = ftrace_preempt_disable();
+        cpu = raw_smp_processor_id();
+        data = tr->data[cpu];
+
+        if (unlikely(atomic_read(&data->disabled)))
+                goto out;
+
+        spin_lock_irqsave(&trace_buf_lock, flags);
+        len = vbin_printf(trace_buf, TRACE_BUF_SIZE, fmt, args);
+
+        if (len > TRACE_BUF_SIZE || len < 0)
+                goto out_unlock;
+
+        size = sizeof(*entry) + sizeof(u32) * len;
+        event = trace_buffer_lock_reserve(tr, TRACE_BPRINTK, size, flags, pc);
+        if (!event)
+                goto out_unlock;
+        entry = ring_buffer_event_data(event);
+        entry->ip                       = ip;
+        entry->fmt                      = fmt;
+
+        memcpy(entry->buf, trace_buf, sizeof(u32) * len);
+        ring_buffer_unlock_commit(tr->buffer, event);
+
+out_unlock:
+        spin_unlock_irqrestore(&trace_buf_lock, flags);
+
+out:
+        ftrace_preempt_enable(resched);
+
+        return len;
+}
+EXPORT_SYMBOL_GPL(trace_vbprintk);
+
 static int trace_panic_handler(struct notifier_block *this,
                                unsigned long event, void *unused)
 {
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 8beff03fda68..0f5077f8f957 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -20,6 +20,7 @@ enum trace_type {
         TRACE_WAKE,
         TRACE_STACK,
         TRACE_PRINT,
+        TRACE_BPRINTK,
         TRACE_SPECIAL,
         TRACE_MMIO_RW,
         TRACE_MMIO_MAP,
@@ -124,6 +125,16 @@ struct print_entry {
         char                    buf[];
 };
 
+struct bprintk_entry {
+        struct trace_entry ent;
+        unsigned long ip;
+        const char *fmt;
+        u32 buf[];
+};
+#ifdef CONFIG_TRACE_BPRINTK
+extern int trace_bprintk_enable;
+#endif
+
 #define TRACE_OLD_SIZE          88
 
 struct trace_field_cont {
@@ -285,6 +296,7 @@ extern void __ftrace_bad_type(void);
                 IF_ASSIGN(var, ent, struct stack_entry, TRACE_STACK);   \
                 IF_ASSIGN(var, ent, struct userstack_entry, TRACE_USER_STACK);\
                 IF_ASSIGN(var, ent, struct print_entry, TRACE_PRINT);   \
+                IF_ASSIGN(var, ent, struct bprintk_entry, TRACE_BPRINTK);\
                 IF_ASSIGN(var, ent, struct special_entry, 0);           \
                 IF_ASSIGN(var, ent, struct trace_mmiotrace_rw,          \
                           TRACE_MMIO_RW);                               \
diff --git a/kernel/trace/trace_bprintk.c b/kernel/trace/trace_bprintk.c
new file mode 100644
index 000000000000..1f8e532c3fb9
--- /dev/null
+++ b/kernel/trace/trace_bprintk.c
@@ -0,0 +1,87 @@
+/*
+ * trace binary printk
+ *
+ * Copyright (C) 2008 Lai Jiangshan <laijs@cn.fujitsu.com>
+ *
+ */
+#include <linux/kernel.h>
+#include <linux/ftrace.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/list.h>
+#include <linux/mutex.h>
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/seq_file.h>
+#include <linux/fs.h>
+#include <linux/marker.h>
+#include <linux/uaccess.h>
+
+#include "trace.h"
+
+/* binary printk basic */
+static DEFINE_MUTEX(btrace_mutex);
+static int btrace_metadata_count;
+
+static inline void lock_btrace(void)
+{
+        mutex_lock(&btrace_mutex);
+}
+
+static inline void unlock_btrace(void)
+{
+        mutex_unlock(&btrace_mutex);
+}
+
+static void get_btrace_metadata(void)
+{
+        lock_btrace();
+        btrace_metadata_count++;
+        unlock_btrace();
+}
+
+static void put_btrace_metadata(void)
+{
+        lock_btrace();
+        btrace_metadata_count--;
+        unlock_btrace();
+}
+
+/* events tracer */
+int trace_bprintk_enable;
+
+static void start_bprintk_trace(struct trace_array *tr)
+{
+        get_btrace_metadata();
+        tracing_reset_online_cpus(tr);
+        trace_bprintk_enable = 1;
+}
+
+static void stop_bprintk_trace(struct trace_array *tr)
+{
+        trace_bprintk_enable = 0;
+        tracing_reset_online_cpus(tr);
+        put_btrace_metadata();
+}
+
+static int init_bprintk_trace(struct trace_array *tr)
+{
+        start_bprintk_trace(tr);
+        return 0;
+}
+
+static struct tracer bprintk_trace __read_mostly =
+{
+        .name        = "events",
+        .init        = init_bprintk_trace,
+        .reset       = stop_bprintk_trace,
+        .start       = start_bprintk_trace,
+        .stop        = stop_bprintk_trace,
+};
+
+static __init int init_bprintk(void)
+{
+        return register_tracer(&bprintk_trace);
+}
+
+device_initcall(init_bprintk);
diff --git a/kernel/trace/trace_output.c b/kernel/trace/trace_output.c
index 306fef84c503..4ab71201862e 100644
--- a/kernel/trace/trace_output.c
+++ b/kernel/trace/trace_output.c
@@ -53,6 +53,26 @@ trace_seq_printf(struct trace_seq *s, const char *fmt, ...)
         return len;
 }
 
+static int
+trace_seq_bprintf(struct trace_seq *s, const char *fmt, const u32 *binary)
+{
+        int len = (PAGE_SIZE - 1) - s->len;
+        int ret;
+
+        if (!len)
+                return 0;
+
+        ret = bstr_printf(s->buffer + s->len, len, fmt, binary);
+
+        /* If we can't write it all, don't bother writing anything */
+        if (ret >= len)
+                return 0;
+
+        s->len += ret;
+
+        return len;
+}
+
 /**
  * trace_seq_puts - trace sequence printing of simple string
  * @s: trace sequence descriptor
@@ -855,6 +875,60 @@ static struct trace_event trace_print_event = {
         .raw            = trace_print_raw,
 };
 
+/* TRACE_BPRINTK */
+static enum print_line_t
+trace_bprintk_print(struct trace_iterator *iter, int flags)
+{
+        struct trace_entry *entry = iter->ent;
+        struct trace_seq *s = &iter->seq;
+        struct bprintk_entry *field;
+
+        trace_assign_type(field, entry);
+
+        if (!seq_print_ip_sym(s, field->ip, flags))
+                goto partial;
+
+        if (!trace_seq_puts(s, ": "))
+                goto partial;
+
+        if (!trace_seq_bprintf(s, field->fmt, field->buf))
+                goto partial;
+
+        return TRACE_TYPE_HANDLED;
+
+ partial:
+        return TRACE_TYPE_PARTIAL_LINE;
+}
+
+static enum print_line_t
+trace_bprintk_raw(struct trace_iterator *iter, int flags)
+{
+        struct trace_entry *entry = iter->ent;
+        struct trace_seq *s = &iter->seq;
+        struct bprintk_entry *field;
+
+        trace_assign_type(field, entry);
+
+        if (!trace_seq_printf(s, ": %lx : ", field->ip))
+                goto partial;
+
+        if (!trace_seq_bprintf(s, field->fmt, field->buf))
+                goto partial;
+
+        return TRACE_TYPE_HANDLED;
+
+ partial:
+        return TRACE_TYPE_PARTIAL_LINE;
+}
+
+static struct trace_event trace_bprintk_event = {
+        .type           = TRACE_BPRINTK,
+        .trace          = trace_bprintk_print,
+        .raw            = trace_bprintk_raw,
+        .hex            = trace_nop_print,
+        .binary         = trace_nop_print,
+};
+
 static struct trace_event *events[] __initdata = {
         &trace_fn_event,
         &trace_ctx_event,
@@ -863,6 +937,7 @@ static struct trace_event *events[] __initdata = {
         &trace_stack_event,
         &trace_user_stack_event,
         &trace_print_event,
+        &trace_bprintk_event,
         NULL
 };