tracing: Use the perf recursion protection from trace event
When we commit a trace to perf, we first check if we are
recursing in the same buffer so that we don't mess-up the buffer
with a recursing trace. But later on, we do the same check from
perf to avoid commit recursion. The recursion check is desired
early before we touch the buffer but we want to do this check
only once.
Then export the recursion protection from perf and use it from
the trace events before submitting a trace.
v2: Put appropriate Reported-by tag
Reported-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Jason Baron <jbaron@redhat.com>
LKML-Reference: <1258864015-10579-1-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
diff --git a/kernel/perf_event.c b/kernel/perf_event.c
index 718fa93..aba82272 100644
--- a/kernel/perf_event.c
+++ b/kernel/perf_event.c
@@ -3880,34 +3880,42 @@
}
}
-static int *perf_swevent_recursion_context(struct perf_cpu_context *cpuctx)
+/*
+ * Must be called with preemption disabled
+ */
+int perf_swevent_get_recursion_context(int **recursion)
{
+ struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
+
if (in_nmi())
- return &cpuctx->recursion[3];
+ *recursion = &cpuctx->recursion[3];
+ else if (in_irq())
+ *recursion = &cpuctx->recursion[2];
+ else if (in_softirq())
+ *recursion = &cpuctx->recursion[1];
+ else
+ *recursion = &cpuctx->recursion[0];
- if (in_irq())
- return &cpuctx->recursion[2];
+ if (**recursion)
+ return -1;
- if (in_softirq())
- return &cpuctx->recursion[1];
+ (**recursion)++;
- return &cpuctx->recursion[0];
+ return 0;
}
-static void do_perf_sw_event(enum perf_type_id type, u32 event_id,
- u64 nr, int nmi,
- struct perf_sample_data *data,
- struct pt_regs *regs)
+void perf_swevent_put_recursion_context(int *recursion)
{
- struct perf_cpu_context *cpuctx = &get_cpu_var(perf_cpu_context);
- int *recursion = perf_swevent_recursion_context(cpuctx);
+ (*recursion)--;
+}
+
+static void __do_perf_sw_event(enum perf_type_id type, u32 event_id,
+ u64 nr, int nmi,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
+{
struct perf_event_context *ctx;
-
- if (*recursion)
- goto out;
-
- (*recursion)++;
- barrier();
+ struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
rcu_read_lock();
perf_swevent_ctx_event(&cpuctx->ctx, type, event_id,
@@ -3920,12 +3928,25 @@
if (ctx)
perf_swevent_ctx_event(ctx, type, event_id, nr, nmi, data, regs);
rcu_read_unlock();
+}
- barrier();
- (*recursion)--;
+static void do_perf_sw_event(enum perf_type_id type, u32 event_id,
+ u64 nr, int nmi,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
+{
+ int *recursion;
+ preempt_disable();
+
+ if (perf_swevent_get_recursion_context(&recursion))
+ goto out;
+
+ __do_perf_sw_event(type, event_id, nr, nmi, data, regs);
+
+ perf_swevent_put_recursion_context(recursion);
out:
- put_cpu_var(perf_cpu_context);
+ preempt_enable();
}
void __perf_sw_event(u32 event_id, u64 nr, int nmi,
@@ -4159,7 +4180,8 @@
if (!regs)
regs = task_pt_regs(current);
- do_perf_sw_event(PERF_TYPE_TRACEPOINT, event_id, count, 1,
+ /* Trace events already protected against recursion */
+ __do_perf_sw_event(PERF_TYPE_TRACEPOINT, event_id, count, 1,
&data, regs);
}
EXPORT_SYMBOL_GPL(perf_tp_event);
diff --git a/kernel/trace/trace_event_profile.c b/kernel/trace/trace_event_profile.c
index e0d351b..d9c60f8 100644
--- a/kernel/trace/trace_event_profile.c
+++ b/kernel/trace/trace_event_profile.c
@@ -9,31 +9,33 @@
#include "trace.h"
-struct perf_trace_buf *perf_trace_buf;
+char *perf_trace_buf;
EXPORT_SYMBOL_GPL(perf_trace_buf);
-struct perf_trace_buf *perf_trace_buf_nmi;
+char *perf_trace_buf_nmi;
EXPORT_SYMBOL_GPL(perf_trace_buf_nmi);
+typedef typeof(char [FTRACE_MAX_PROFILE_SIZE]) perf_trace_t ;
+
/* Count the events in use (per event id, not per instance) */
static int total_profile_count;
static int ftrace_profile_enable_event(struct ftrace_event_call *event)
{
- struct perf_trace_buf *buf;
+ char *buf;
int ret = -ENOMEM;
if (atomic_inc_return(&event->profile_count))
return 0;
if (!total_profile_count) {
- buf = alloc_percpu(struct perf_trace_buf);
+ buf = (char *)alloc_percpu(perf_trace_t);
if (!buf)
goto fail_buf;
rcu_assign_pointer(perf_trace_buf, buf);
- buf = alloc_percpu(struct perf_trace_buf);
+ buf = (char *)alloc_percpu(perf_trace_t);
if (!buf)
goto fail_buf_nmi;
@@ -79,7 +81,7 @@
static void ftrace_profile_disable_event(struct ftrace_event_call *event)
{
- struct perf_trace_buf *buf, *nmi_buf;
+ char *buf, *nmi_buf;
if (!atomic_add_negative(-1, &event->profile_count))
return;
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index 3696476..22e6f68 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -1208,11 +1208,12 @@
struct trace_probe *tp = container_of(kp, struct trace_probe, rp.kp);
struct ftrace_event_call *call = &tp->call;
struct kprobe_trace_entry *entry;
- struct perf_trace_buf *trace_buf;
struct trace_entry *ent;
int size, __size, i, pc, __cpu;
unsigned long irq_flags;
+ char *trace_buf;
char *raw_data;
+ int *recursion;
pc = preempt_count();
__size = SIZEOF_KPROBE_TRACE_ENTRY(tp->nr_args);
@@ -1227,6 +1228,10 @@
* This also protects the rcu read side
*/
local_irq_save(irq_flags);
+
+ if (perf_swevent_get_recursion_context(&recursion))
+ goto end_recursion;
+
__cpu = smp_processor_id();
if (in_nmi())
@@ -1237,18 +1242,7 @@
if (!trace_buf)
goto end;
- trace_buf = per_cpu_ptr(trace_buf, __cpu);
-
- if (trace_buf->recursion++)
- goto end_recursion;
-
- /*
- * Make recursion update visible before entering perf_tp_event
- * so that we protect from perf recursions.
- */
- barrier();
-
- raw_data = trace_buf->buf;
+ raw_data = per_cpu_ptr(trace_buf, __cpu);
/* Zero dead bytes from alignment to avoid buffer leak to userspace */
*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
@@ -1263,9 +1257,9 @@
entry->args[i] = call_fetch(&tp->args[i].fetch, regs);
perf_tp_event(call->id, entry->ip, 1, entry, size);
-end_recursion:
- trace_buf->recursion--;
end:
+ perf_swevent_put_recursion_context(recursion);
+end_recursion:
local_irq_restore(irq_flags);
return 0;
@@ -1278,10 +1272,11 @@
struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
struct ftrace_event_call *call = &tp->call;
struct kretprobe_trace_entry *entry;
- struct perf_trace_buf *trace_buf;
struct trace_entry *ent;
int size, __size, i, pc, __cpu;
unsigned long irq_flags;
+ char *trace_buf;
+ int *recursion;
char *raw_data;
pc = preempt_count();
@@ -1297,6 +1292,10 @@
* This also protects the rcu read side
*/
local_irq_save(irq_flags);
+
+ if (perf_swevent_get_recursion_context(&recursion))
+ goto end_recursion;
+
__cpu = smp_processor_id();
if (in_nmi())
@@ -1307,18 +1306,7 @@
if (!trace_buf)
goto end;
- trace_buf = per_cpu_ptr(trace_buf, __cpu);
-
- if (trace_buf->recursion++)
- goto end_recursion;
-
- /*
- * Make recursion update visible before entering perf_tp_event
- * so that we protect from perf recursions.
- */
- barrier();
-
- raw_data = trace_buf->buf;
+ raw_data = per_cpu_ptr(trace_buf, __cpu);
/* Zero dead bytes from alignment to avoid buffer leak to userspace */
*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
@@ -1334,9 +1322,9 @@
entry->args[i] = call_fetch(&tp->args[i].fetch, regs);
perf_tp_event(call->id, entry->ret_ip, 1, entry, size);
-end_recursion:
- trace_buf->recursion--;
end:
+ perf_swevent_put_recursion_context(recursion);
+end_recursion:
local_irq_restore(irq_flags);
return 0;
diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c
index 51213b0..0bb9348 100644
--- a/kernel/trace/trace_syscalls.c
+++ b/kernel/trace/trace_syscalls.c
@@ -477,10 +477,11 @@
static void prof_syscall_enter(struct pt_regs *regs, long id)
{
struct syscall_metadata *sys_data;
- struct perf_trace_buf *trace_buf;
struct syscall_trace_enter *rec;
unsigned long flags;
+ char *trace_buf;
char *raw_data;
+ int *recursion;
int syscall_nr;
int size;
int cpu;
@@ -505,6 +506,9 @@
/* Protect the per cpu buffer, begin the rcu read side */
local_irq_save(flags);
+ if (perf_swevent_get_recursion_context(&recursion))
+ goto end_recursion;
+
cpu = smp_processor_id();
if (in_nmi())
@@ -515,18 +519,7 @@
if (!trace_buf)
goto end;
- trace_buf = per_cpu_ptr(trace_buf, cpu);
-
- if (trace_buf->recursion++)
- goto end_recursion;
-
- /*
- * Make recursion update visible before entering perf_tp_event
- * so that we protect from perf recursions.
- */
- barrier();
-
- raw_data = trace_buf->buf;
+ raw_data = per_cpu_ptr(trace_buf, cpu);
/* zero the dead bytes from align to not leak stack to user */
*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
@@ -539,9 +532,9 @@
(unsigned long *)&rec->args);
perf_tp_event(sys_data->enter_id, 0, 1, rec, size);
-end_recursion:
- trace_buf->recursion--;
end:
+ perf_swevent_put_recursion_context(recursion);
+end_recursion:
local_irq_restore(flags);
}
@@ -588,10 +581,11 @@
{
struct syscall_metadata *sys_data;
struct syscall_trace_exit *rec;
- struct perf_trace_buf *trace_buf;
unsigned long flags;
int syscall_nr;
+ char *trace_buf;
char *raw_data;
+ int *recursion;
int size;
int cpu;
@@ -617,6 +611,10 @@
/* Protect the per cpu buffer, begin the rcu read side */
local_irq_save(flags);
+
+ if (perf_swevent_get_recursion_context(&recursion))
+ goto end_recursion;
+
cpu = smp_processor_id();
if (in_nmi())
@@ -627,18 +625,7 @@
if (!trace_buf)
goto end;
- trace_buf = per_cpu_ptr(trace_buf, cpu);
-
- if (trace_buf->recursion++)
- goto end_recursion;
-
- /*
- * Make recursion update visible before entering perf_tp_event
- * so that we protect from perf recursions.
- */
- barrier();
-
- raw_data = trace_buf->buf;
+ raw_data = per_cpu_ptr(trace_buf, cpu);
/* zero the dead bytes from align to not leak stack to user */
*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
@@ -652,9 +639,9 @@
perf_tp_event(sys_data->exit_id, 0, 1, rec, size);
-end_recursion:
- trace_buf->recursion--;
end:
+ perf_swevent_put_recursion_context(recursion);
+end_recursion:
local_irq_restore(flags);
}
