* [GIT PULL] perf events updates
@ 2010-01-29 2:07 Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 1/5] perf: Fix inconsistency between IP and callchain sampling Frederic Weisbecker
` (5 more replies)
0 siblings, 6 replies; 8+ messages in thread
From: Frederic Weisbecker @ 2010-01-29 2:07 UTC (permalink / raw)
To: Ingo Molnar
Cc: LKML, Frederic Weisbecker, K . Prasad, Xiao Guangrong,
Paul Mackerras, Peter Zijlstra, Anton Blanchard
Ingo,
Please pull the perf/core branch that can be found at:
git://git.kernel.org/pub/scm/linux/kernel/git/frederic/random-tracing.git
perf/core
Thanks,
Frederic
---
K.Prasad (2):
x86/debug: Clear reserved bits of DR6 in do_debug()
x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler
Xiao Guangrong (2):
perf: Factorize trace events raw sample buffer operations
tracing/kprobe: Cleanup unused return value of tracing functions
Anton Blanchard (1):
perf: Fix inconsistency between IP and callchain sampling
arch/powerpc/kernel/perf_callchain.c | 3 -
arch/sh/kernel/perf_callchain.c | 3 -
arch/x86/include/asm/debugreg.h | 3 +
arch/x86/kernel/cpu/perf_event.c | 3 -
arch/x86/kernel/hw_breakpoint.c | 10 ++-
arch/x86/kernel/traps.c | 3 +
include/linux/ftrace_event.h | 18 +++++-
include/trace/ftrace.h | 48 ++-------------
kernel/trace/trace_event_profile.c | 52 +++++++++++++++--
kernel/trace/trace_kprobe.c | 109 ++++++----------------------------
kernel/trace/trace_syscalls.c | 71 +++-------------------
11 files changed, 109 insertions(+), 214 deletions(-)
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 1/5] perf: Fix inconsistency between IP and callchain sampling
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
@ 2010-01-29 2:07 ` Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 2/5] perf: Factorize trace events raw sample buffer operations Frederic Weisbecker
` (4 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Frederic Weisbecker @ 2010-01-29 2:07 UTC (permalink / raw)
To: Ingo Molnar
Cc: LKML, Anton Blanchard, Peter Zijlstra, Paul Mackerras,
Ingo Molnar, Benjamin Herrenschmidt, Paul Mundt,
Frederic Weisbecker
From: Anton Blanchard <anton@samba.org>
When running perf across all cpus with backtracing (-a -g), sometimes we
get samples without associated backtraces:
23.44% init [kernel] [k] restore
11.46% init eeba0c [k] 0x00000000eeba0c
6.77% swapper [kernel] [k] .perf_ctx_adjust_freq
5.73% init [kernel] [k] .__trace_hcall_entry
4.69% perf libc-2.9.so [.] 0x0000000006bb8c
|
|--11.11%-- 0xfffa941bbbc
It turns out the backtrace code has a check for the idle task and the IP
sampling does not. This creates problems when profiling an interrupt
heavy workload (in my case 10Gbit ethernet) since we get no backtraces
for interrupts received while idle (ie most of the workload).
Right now x86 and sh check that current is not NULL, which should never
happen so remove that too.
Idle task's exclusion must be performed from the core code, on top
of perf_event_attr:exclude_idle.
Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mundt <lethal@linux-sh.org>
LKML-Reference: <20100118054707.GT12666@kryten>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
---
arch/powerpc/kernel/perf_callchain.c | 3 ---
arch/sh/kernel/perf_callchain.c | 3 ---
arch/x86/kernel/cpu/perf_event.c | 3 ---
3 files changed, 0 insertions(+), 9 deletions(-)
diff --git a/arch/powerpc/kernel/perf_callchain.c b/arch/powerpc/kernel/perf_callchain.c
index a3c11ca..95ad9da 100644
--- a/arch/powerpc/kernel/perf_callchain.c
+++ b/arch/powerpc/kernel/perf_callchain.c
@@ -495,9 +495,6 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs)
entry->nr = 0;
- if (current->pid == 0) /* idle task? */
- return entry;
-
if (!user_mode(regs)) {
perf_callchain_kernel(regs, entry);
if (current->mm)
diff --git a/arch/sh/kernel/perf_callchain.c b/arch/sh/kernel/perf_callchain.c
index 24ea837..a9dd3ab 100644
--- a/arch/sh/kernel/perf_callchain.c
+++ b/arch/sh/kernel/perf_callchain.c
@@ -68,9 +68,6 @@ perf_do_callchain(struct pt_regs *regs, struct perf_callchain_entry *entry)
is_user = user_mode(regs);
- if (!current || current->pid == 0)
- return;
-
if (is_user && current->state != TASK_RUNNING)
return;
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index b1bb8c5..ed1998b 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -2425,9 +2425,6 @@ perf_do_callchain(struct pt_regs *regs, struct perf_callchain_entry *entry)
is_user = user_mode(regs);
- if (!current || current->pid == 0)
- return;
-
if (is_user && current->state != TASK_RUNNING)
return;
--
1.6.2.3
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 2/5] perf: Factorize trace events raw sample buffer operations
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 1/5] perf: Fix inconsistency between IP and callchain sampling Frederic Weisbecker
@ 2010-01-29 2:07 ` Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 3/5] tracing/kprobe: Cleanup unused return value of tracing functions Frederic Weisbecker
` (3 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Frederic Weisbecker @ 2010-01-29 2:07 UTC (permalink / raw)
To: Ingo Molnar
Cc: LKML, Xiao Guangrong, Ingo Molnar, Steven Rostedt,
Paul Mackerras, Jason Baron, Peter Zijlstra, Frederic Weisbecker
From: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Introduce ftrace_perf_buf_prepare() and ftrace_perf_buf_submit() to
gather the common code that operates on raw events sampling buffer.
This cleans up redundant code between regular trace events, syscall
events and kprobe events.
Changelog v1->v2:
- Rename function name as per Masami and Frederic's suggestion
- Add __kprobes for ftrace_perf_buf_prepare() and make
ftrace_perf_buf_submit() inline as per Masami's suggestion
- Export ftrace_perf_buf_prepare since modules will use it
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Baron <jbaron@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <4B60E92D.9000808@cn.fujitsu.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
---
include/linux/ftrace_event.h | 18 ++++++-
include/trace/ftrace.h | 48 +++-----------------
kernel/trace/trace_event_profile.c | 52 +++++++++++++++++++--
kernel/trace/trace_kprobe.c | 86 ++++-------------------------------
kernel/trace/trace_syscalls.c | 71 ++++-------------------------
5 files changed, 88 insertions(+), 187 deletions(-)
diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
index 0a09e75..cd95919 100644
--- a/include/linux/ftrace_event.h
+++ b/include/linux/ftrace_event.h
@@ -5,6 +5,7 @@
#include <linux/trace_seq.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
+#include <linux/perf_event.h>
struct trace_array;
struct tracer;
@@ -138,9 +139,6 @@ struct ftrace_event_call {
#define FTRACE_MAX_PROFILE_SIZE 2048
-extern char *perf_trace_buf;
-extern char *perf_trace_buf_nmi;
-
#define MAX_FILTER_PRED 32
#define MAX_FILTER_STR_VAL 256 /* Should handle KSYM_SYMBOL_LEN */
@@ -195,6 +193,20 @@ extern void ftrace_profile_disable(int event_id);
extern int ftrace_profile_set_filter(struct perf_event *event, int event_id,
char *filter_str);
extern void ftrace_profile_free_filter(struct perf_event *event);
+extern void *
+ftrace_perf_buf_prepare(int size, unsigned short type, int *rctxp,
+ unsigned long *irq_flags);
+
+static inline void
+ftrace_perf_buf_submit(void *raw_data, int size, int rctx, u64 addr,
+ u64 count, unsigned long irq_flags)
+{
+ struct trace_entry *entry = raw_data;
+
+ perf_tp_event(entry->type, addr, count, raw_data, size);
+ perf_swevent_put_recursion_context(rctx);
+ local_irq_restore(irq_flags);
+}
#endif
#endif /* _LINUX_FTRACE_EVENT_H */
diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h
index 4a46a60..f2c09e4 100644
--- a/include/trace/ftrace.h
+++ b/include/trace/ftrace.h
@@ -850,22 +850,12 @@ ftrace_profile_templ_##call(struct ftrace_event_call *event_call, \
proto) \
{ \
struct ftrace_data_offsets_##call __maybe_unused __data_offsets;\
- extern int perf_swevent_get_recursion_context(void); \
- extern void perf_swevent_put_recursion_context(int rctx); \
- extern void perf_tp_event(int, u64, u64, void *, int); \
struct ftrace_raw_##call *entry; \
u64 __addr = 0, __count = 1; \
unsigned long irq_flags; \
- struct trace_entry *ent; \
int __entry_size; \
int __data_size; \
- char *trace_buf; \
- char *raw_data; \
- int __cpu; \
int rctx; \
- int pc; \
- \
- pc = preempt_count(); \
\
__data_size = ftrace_get_offsets_##call(&__data_offsets, args); \
__entry_size = ALIGN(__data_size + sizeof(*entry) + sizeof(u32),\
@@ -875,42 +865,16 @@ ftrace_profile_templ_##call(struct ftrace_event_call *event_call, \
if (WARN_ONCE(__entry_size > FTRACE_MAX_PROFILE_SIZE, \
"profile buffer not large enough")) \
return; \
- \
- local_irq_save(irq_flags); \
- \
- rctx = perf_swevent_get_recursion_context(); \
- if (rctx < 0) \
- goto end_recursion; \
- \
- __cpu = smp_processor_id(); \
- \
- if (in_nmi()) \
- trace_buf = rcu_dereference(perf_trace_buf_nmi); \
- else \
- trace_buf = rcu_dereference(perf_trace_buf); \
- \
- if (!trace_buf) \
- goto end; \
- \
- raw_data = per_cpu_ptr(trace_buf, __cpu); \
- \
- *(u64 *)(&raw_data[__entry_size - sizeof(u64)]) = 0ULL; \
- entry = (struct ftrace_raw_##call *)raw_data; \
- ent = &entry->ent; \
- tracing_generic_entry_update(ent, irq_flags, pc); \
- ent->type = event_call->id; \
- \
+ entry = (struct ftrace_raw_##call *)ftrace_perf_buf_prepare( \
+ __entry_size, event_call->id, &rctx, &irq_flags); \
+ if (!entry) \
+ return; \
tstruct \
\
{ assign; } \
\
- perf_tp_event(event_call->id, __addr, __count, entry, \
- __entry_size); \
- \
-end: \
- perf_swevent_put_recursion_context(rctx); \
-end_recursion: \
- local_irq_restore(irq_flags); \
+ ftrace_perf_buf_submit(entry, __entry_size, rctx, __addr, \
+ __count, irq_flags); \
}
#undef DEFINE_EVENT
diff --git a/kernel/trace/trace_event_profile.c b/kernel/trace/trace_event_profile.c
index 9e25573..f0d6930 100644
--- a/kernel/trace/trace_event_profile.c
+++ b/kernel/trace/trace_event_profile.c
@@ -6,14 +6,12 @@
*/
#include <linux/module.h>
+#include <linux/kprobes.h>
#include "trace.h"
-char *perf_trace_buf;
-EXPORT_SYMBOL_GPL(perf_trace_buf);
-
-char *perf_trace_buf_nmi;
-EXPORT_SYMBOL_GPL(perf_trace_buf_nmi);
+static char *perf_trace_buf;
+static char *perf_trace_buf_nmi;
typedef typeof(char [FTRACE_MAX_PROFILE_SIZE]) perf_trace_t ;
@@ -120,3 +118,47 @@ void ftrace_profile_disable(int event_id)
}
mutex_unlock(&event_mutex);
}
+
+__kprobes void *ftrace_perf_buf_prepare(int size, unsigned short type,
+ int *rctxp, unsigned long *irq_flags)
+{
+ struct trace_entry *entry;
+ char *trace_buf, *raw_data;
+ int pc, cpu;
+
+ pc = preempt_count();
+
+ /* Protect the per cpu buffer, begin the rcu read side */
+ local_irq_save(*irq_flags);
+
+ *rctxp = perf_swevent_get_recursion_context();
+ if (*rctxp < 0)
+ goto err_recursion;
+
+ cpu = smp_processor_id();
+
+ if (in_nmi())
+ trace_buf = rcu_dereference(perf_trace_buf_nmi);
+ else
+ trace_buf = rcu_dereference(perf_trace_buf);
+
+ if (!trace_buf)
+ goto err;
+
+ 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;
+
+ entry = (struct trace_entry *)raw_data;
+ tracing_generic_entry_update(entry, *irq_flags, pc);
+ entry->type = type;
+
+ return raw_data;
+err:
+ perf_swevent_put_recursion_context(*rctxp);
+err_recursion:
+ local_irq_restore(*irq_flags);
+ return NULL;
+}
+EXPORT_SYMBOL_GPL(ftrace_perf_buf_prepare);
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index d6266ca..2e28ee3 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -1243,14 +1243,10 @@ static __kprobes int kprobe_profile_func(struct kprobe *kp,
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 trace_entry *ent;
- int size, __size, i, pc, __cpu;
+ int size, __size, i;
unsigned long irq_flags;
- char *trace_buf;
- char *raw_data;
int rctx;
- pc = preempt_count();
__size = SIZEOF_KPROBE_TRACE_ENTRY(tp->nr_args);
size = ALIGN(__size + sizeof(u32), sizeof(u64));
size -= sizeof(u32);
@@ -1258,45 +1254,16 @@ static __kprobes int kprobe_profile_func(struct kprobe *kp,
"profile buffer not large enough"))
return 0;
- /*
- * Protect the non nmi buffer
- * This also protects the rcu read side
- */
- local_irq_save(irq_flags);
-
- rctx = perf_swevent_get_recursion_context();
- if (rctx < 0)
- goto end_recursion;
-
- __cpu = smp_processor_id();
-
- if (in_nmi())
- trace_buf = rcu_dereference(perf_trace_buf_nmi);
- else
- trace_buf = rcu_dereference(perf_trace_buf);
-
- if (!trace_buf)
- goto end;
-
- 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;
- entry = (struct kprobe_trace_entry *)raw_data;
- ent = &entry->ent;
+ entry = ftrace_perf_buf_prepare(size, call->id, &rctx, &irq_flags);
+ if (!entry)
+ return 0;
- tracing_generic_entry_update(ent, irq_flags, pc);
- ent->type = call->id;
entry->nargs = tp->nr_args;
entry->ip = (unsigned long)kp->addr;
for (i = 0; i < tp->nr_args; i++)
entry->args[i] = call_fetch(&tp->args[i].fetch, regs);
- perf_tp_event(call->id, entry->ip, 1, entry, size);
-end:
- perf_swevent_put_recursion_context(rctx);
-end_recursion:
- local_irq_restore(irq_flags);
+ ftrace_perf_buf_submit(entry, size, rctx, entry->ip, 1, irq_flags);
return 0;
}
@@ -1308,14 +1275,10 @@ static __kprobes int kretprobe_profile_func(struct kretprobe_instance *ri,
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 trace_entry *ent;
- int size, __size, i, pc, __cpu;
+ int size, __size, i;
unsigned long irq_flags;
- char *trace_buf;
- char *raw_data;
int rctx;
- pc = preempt_count();
__size = SIZEOF_KRETPROBE_TRACE_ENTRY(tp->nr_args);
size = ALIGN(__size + sizeof(u32), sizeof(u64));
size -= sizeof(u32);
@@ -1323,46 +1286,17 @@ static __kprobes int kretprobe_profile_func(struct kretprobe_instance *ri,
"profile buffer not large enough"))
return 0;
- /*
- * Protect the non nmi buffer
- * This also protects the rcu read side
- */
- local_irq_save(irq_flags);
-
- rctx = perf_swevent_get_recursion_context();
- if (rctx < 0)
- goto end_recursion;
-
- __cpu = smp_processor_id();
-
- if (in_nmi())
- trace_buf = rcu_dereference(perf_trace_buf_nmi);
- else
- trace_buf = rcu_dereference(perf_trace_buf);
-
- if (!trace_buf)
- goto end;
-
- 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;
- entry = (struct kretprobe_trace_entry *)raw_data;
- ent = &entry->ent;
+ entry = ftrace_perf_buf_prepare(size, call->id, &rctx, &irq_flags);
+ if (!entry)
+ return 0;
- tracing_generic_entry_update(ent, irq_flags, pc);
- ent->type = call->id;
entry->nargs = tp->nr_args;
entry->func = (unsigned long)tp->rp.kp.addr;
entry->ret_ip = (unsigned long)ri->ret_addr;
for (i = 0; i < tp->nr_args; i++)
entry->args[i] = call_fetch(&tp->args[i].fetch, regs);
- perf_tp_event(call->id, entry->ret_ip, 1, entry, size);
-end:
- perf_swevent_put_recursion_context(rctx);
-end_recursion:
- local_irq_restore(irq_flags);
+ ftrace_perf_buf_submit(entry, size, rctx, entry->ret_ip, 1, irq_flags);
return 0;
}
diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c
index f694f66..4e332b9 100644
--- a/kernel/trace/trace_syscalls.c
+++ b/kernel/trace/trace_syscalls.c
@@ -433,12 +433,9 @@ static void prof_syscall_enter(struct pt_regs *regs, long id)
struct syscall_metadata *sys_data;
struct syscall_trace_enter *rec;
unsigned long flags;
- char *trace_buf;
- char *raw_data;
int syscall_nr;
int rctx;
int size;
- int cpu;
syscall_nr = syscall_get_nr(current, regs);
if (!test_bit(syscall_nr, enabled_prof_enter_syscalls))
@@ -457,37 +454,15 @@ static void prof_syscall_enter(struct pt_regs *regs, long id)
"profile buffer not large enough"))
return;
- /* Protect the per cpu buffer, begin the rcu read side */
- local_irq_save(flags);
-
- rctx = perf_swevent_get_recursion_context();
- if (rctx < 0)
- goto end_recursion;
-
- cpu = smp_processor_id();
-
- trace_buf = rcu_dereference(perf_trace_buf);
-
- if (!trace_buf)
- goto end;
-
- 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;
+ rec = (struct syscall_trace_enter *)ftrace_perf_buf_prepare(size,
+ sys_data->enter_event->id, &rctx, &flags);
+ if (!rec)
+ return;
- rec = (struct syscall_trace_enter *) raw_data;
- tracing_generic_entry_update(&rec->ent, 0, 0);
- rec->ent.type = sys_data->enter_event->id;
rec->nr = syscall_nr;
syscall_get_arguments(current, regs, 0, sys_data->nb_args,
(unsigned long *)&rec->args);
- perf_tp_event(sys_data->enter_event->id, 0, 1, rec, size);
-
-end:
- perf_swevent_put_recursion_context(rctx);
-end_recursion:
- local_irq_restore(flags);
+ ftrace_perf_buf_submit(rec, size, rctx, 0, 1, flags);
}
int prof_sysenter_enable(struct ftrace_event_call *call)
@@ -531,11 +506,8 @@ static void prof_syscall_exit(struct pt_regs *regs, long ret)
struct syscall_trace_exit *rec;
unsigned long flags;
int syscall_nr;
- char *trace_buf;
- char *raw_data;
int rctx;
int size;
- int cpu;
syscall_nr = syscall_get_nr(current, regs);
if (!test_bit(syscall_nr, enabled_prof_exit_syscalls))
@@ -557,38 +529,15 @@ static void prof_syscall_exit(struct pt_regs *regs, long ret)
"exit event has grown above profile buffer size"))
return;
- /* Protect the per cpu buffer, begin the rcu read side */
- local_irq_save(flags);
-
- rctx = perf_swevent_get_recursion_context();
- if (rctx < 0)
- goto end_recursion;
-
- cpu = smp_processor_id();
-
- trace_buf = rcu_dereference(perf_trace_buf);
-
- if (!trace_buf)
- goto end;
-
- 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;
-
- rec = (struct syscall_trace_exit *)raw_data;
+ rec = (struct syscall_trace_exit *)ftrace_perf_buf_prepare(size,
+ sys_data->exit_event->id, &rctx, &flags);
+ if (!rec)
+ return;
- tracing_generic_entry_update(&rec->ent, 0, 0);
- rec->ent.type = sys_data->exit_event->id;
rec->nr = syscall_nr;
rec->ret = syscall_get_return_value(current, regs);
- perf_tp_event(sys_data->exit_event->id, 0, 1, rec, size);
-
-end:
- perf_swevent_put_recursion_context(rctx);
-end_recursion:
- local_irq_restore(flags);
+ ftrace_perf_buf_submit(rec, size, rctx, 0, 1, flags);
}
int prof_sysexit_enable(struct ftrace_event_call *call)
--
1.6.2.3
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 3/5] tracing/kprobe: Cleanup unused return value of tracing functions
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 1/5] perf: Fix inconsistency between IP and callchain sampling Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 2/5] perf: Factorize trace events raw sample buffer operations Frederic Weisbecker
@ 2010-01-29 2:07 ` Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 4/5] x86/debug: Clear reserved bits of DR6 in do_debug() Frederic Weisbecker
` (2 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Frederic Weisbecker @ 2010-01-29 2:07 UTC (permalink / raw)
To: Ingo Molnar
Cc: LKML, Xiao Guangrong, Steven Rostedt, Ingo Molnar,
Paul Mackerras, Jason Baron, Peter Zijlstra, Frederic Weisbecker
From: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
The return values of the kprobe's tracing functions are meaningless,
lets remove these.
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Acked-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Baron <jbaron@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <4B60E9A3.2040505@cn.fujitsu.com>
[fweisbec@gmail: whitespace fixes, drop useless void returns in end
of functions]
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
---
kernel/trace/trace_kprobe.c | 27 ++++++++++-----------------
1 files changed, 10 insertions(+), 17 deletions(-)
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index 2e28ee3..6178abf 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -942,7 +942,7 @@ static const struct file_operations kprobe_profile_ops = {
};
/* Kprobe handler */
-static __kprobes int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
+static __kprobes void kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
{
struct trace_probe *tp = container_of(kp, struct trace_probe, rp.kp);
struct kprobe_trace_entry *entry;
@@ -962,7 +962,7 @@ static __kprobes int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
event = trace_current_buffer_lock_reserve(&buffer, call->id, size,
irq_flags, pc);
if (!event)
- return 0;
+ return;
entry = ring_buffer_event_data(event);
entry->nargs = tp->nr_args;
@@ -972,11 +972,10 @@ static __kprobes int kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs)
if (!filter_current_check_discard(buffer, call, entry, event))
trace_nowake_buffer_unlock_commit(buffer, event, irq_flags, pc);
- return 0;
}
/* Kretprobe handler */
-static __kprobes int kretprobe_trace_func(struct kretprobe_instance *ri,
+static __kprobes void kretprobe_trace_func(struct kretprobe_instance *ri,
struct pt_regs *regs)
{
struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
@@ -995,7 +994,7 @@ static __kprobes int kretprobe_trace_func(struct kretprobe_instance *ri,
event = trace_current_buffer_lock_reserve(&buffer, call->id, size,
irq_flags, pc);
if (!event)
- return 0;
+ return;
entry = ring_buffer_event_data(event);
entry->nargs = tp->nr_args;
@@ -1006,8 +1005,6 @@ static __kprobes int kretprobe_trace_func(struct kretprobe_instance *ri,
if (!filter_current_check_discard(buffer, call, entry, event))
trace_nowake_buffer_unlock_commit(buffer, event, irq_flags, pc);
-
- return 0;
}
/* Event entry printers */
@@ -1237,7 +1234,7 @@ static int kretprobe_event_show_format(struct ftrace_event_call *call,
#ifdef CONFIG_PERF_EVENTS
/* Kprobe profile handler */
-static __kprobes int kprobe_profile_func(struct kprobe *kp,
+static __kprobes void kprobe_profile_func(struct kprobe *kp,
struct pt_regs *regs)
{
struct trace_probe *tp = container_of(kp, struct trace_probe, rp.kp);
@@ -1252,11 +1249,11 @@ static __kprobes int kprobe_profile_func(struct kprobe *kp,
size -= sizeof(u32);
if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
"profile buffer not large enough"))
- return 0;
+ return;
entry = ftrace_perf_buf_prepare(size, call->id, &rctx, &irq_flags);
if (!entry)
- return 0;
+ return;
entry->nargs = tp->nr_args;
entry->ip = (unsigned long)kp->addr;
@@ -1264,12 +1261,10 @@ static __kprobes int kprobe_profile_func(struct kprobe *kp,
entry->args[i] = call_fetch(&tp->args[i].fetch, regs);
ftrace_perf_buf_submit(entry, size, rctx, entry->ip, 1, irq_flags);
-
- return 0;
}
/* Kretprobe profile handler */
-static __kprobes int kretprobe_profile_func(struct kretprobe_instance *ri,
+static __kprobes void kretprobe_profile_func(struct kretprobe_instance *ri,
struct pt_regs *regs)
{
struct trace_probe *tp = container_of(ri->rp, struct trace_probe, rp);
@@ -1284,11 +1279,11 @@ static __kprobes int kretprobe_profile_func(struct kretprobe_instance *ri,
size -= sizeof(u32);
if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
"profile buffer not large enough"))
- return 0;
+ return;
entry = ftrace_perf_buf_prepare(size, call->id, &rctx, &irq_flags);
if (!entry)
- return 0;
+ return;
entry->nargs = tp->nr_args;
entry->func = (unsigned long)tp->rp.kp.addr;
@@ -1297,8 +1292,6 @@ static __kprobes int kretprobe_profile_func(struct kretprobe_instance *ri,
entry->args[i] = call_fetch(&tp->args[i].fetch, regs);
ftrace_perf_buf_submit(entry, size, rctx, entry->ret_ip, 1, irq_flags);
-
- return 0;
}
static int probe_profile_enable(struct ftrace_event_call *call)
--
1.6.2.3
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 4/5] x86/debug: Clear reserved bits of DR6 in do_debug()
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
` (2 preceding siblings ...)
2010-01-29 2:07 ` [PATCH 3/5] tracing/kprobe: Cleanup unused return value of tracing functions Frederic Weisbecker
@ 2010-01-29 2:07 ` Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 5/5] x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler Frederic Weisbecker
2010-01-29 7:06 ` [GIT PULL] perf events updates Ingo Molnar
5 siblings, 0 replies; 8+ messages in thread
From: Frederic Weisbecker @ 2010-01-29 2:07 UTC (permalink / raw)
To: Ingo Molnar
Cc: LKML, K.Prasad, Roland McGrath, Jan Kiszka, Alan Stern,
Ingo Molnar, Frederic Weisbecker
From: K.Prasad <prasad@linux.vnet.ibm.com>
Clear the reserved bits from the stored copy of debug status
register (DR6).
This will help easy bitwise operations such as quick testing
of a debug event origin.
Signed-off-by: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Ingo Molnar <mingo@elte.hu>
LKML-Reference: <20100128111401.GB13935@in.ibm.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
---
arch/x86/include/asm/debugreg.h | 3 +++
arch/x86/kernel/traps.c | 3 +++
2 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/arch/x86/include/asm/debugreg.h b/arch/x86/include/asm/debugreg.h
index 8240f76..b81002f 100644
--- a/arch/x86/include/asm/debugreg.h
+++ b/arch/x86/include/asm/debugreg.h
@@ -14,6 +14,9 @@
which debugging register was responsible for the trap. The other bits
are either reserved or not of interest to us. */
+/* Define reserved bits in DR6 which are always set to 1 */
+#define DR6_RESERVED (0xFFFF0FF0)
+
#define DR_TRAP0 (0x1) /* db0 */
#define DR_TRAP1 (0x2) /* db1 */
#define DR_TRAP2 (0x4) /* db2 */
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 3339917..1168e44 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -534,6 +534,9 @@ dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code)
get_debugreg(dr6, 6);
+ /* Filter out all the reserved bits which are preset to 1 */
+ dr6 &= ~DR6_RESERVED;
+
/* Catch kmemcheck conditions first of all! */
if ((dr6 & DR_STEP) && kmemcheck_trap(regs))
return;
--
1.6.2.3
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 5/5] x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
` (3 preceding siblings ...)
2010-01-29 2:07 ` [PATCH 4/5] x86/debug: Clear reserved bits of DR6 in do_debug() Frederic Weisbecker
@ 2010-01-29 2:07 ` Frederic Weisbecker
2010-01-29 7:06 ` [GIT PULL] perf events updates Ingo Molnar
5 siblings, 0 replies; 8+ messages in thread
From: Frederic Weisbecker @ 2010-01-29 2:07 UTC (permalink / raw)
To: Ingo Molnar
Cc: LKML, K.Prasad, Ingo Molnar, Roland McGrath, Alan Stern,
Jan Kiszka, Frederic Weisbecker
From: K.Prasad <prasad@linux.vnet.ibm.com>
Processing of debug exceptions in do_debug() can stop if it
originated from a hw-breakpoint exception by returning NOTIFY_STOP
in most cases.
But for certain cases such as:
a) user-space breakpoints with pending SIGTRAP signal delivery (as
in the case of ptrace induced breakpoints).
b) exceptions due to other causes than breakpoints
We will continue to process the exception by returning NOTIFY_DONE.
Signed-off-by: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland McGrath <roland@redhat.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Jan Kiszka <jan.kiszka@siemens.com>
LKML-Reference: <20100128111415.GC13935@in.ibm.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
---
arch/x86/kernel/hw_breakpoint.c | 10 +++++++---
1 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/arch/x86/kernel/hw_breakpoint.c b/arch/x86/kernel/hw_breakpoint.c
index 05d5fec..ae90b47 100644
--- a/arch/x86/kernel/hw_breakpoint.c
+++ b/arch/x86/kernel/hw_breakpoint.c
@@ -502,8 +502,6 @@ static int __kprobes hw_breakpoint_handler(struct die_args *args)
rcu_read_lock();
bp = per_cpu(bp_per_reg[i], cpu);
- if (bp)
- rc = NOTIFY_DONE;
/*
* Reset the 'i'th TRAP bit in dr6 to denote completion of
* exception handling
@@ -522,7 +520,13 @@ static int __kprobes hw_breakpoint_handler(struct die_args *args)
rcu_read_unlock();
}
- if (dr6 & (~DR_TRAP_BITS))
+ /*
+ * Further processing in do_debug() is needed for a) user-space
+ * breakpoints (to generate signals) and b) when the system has
+ * taken exception due to multiple causes
+ */
+ if ((current->thread.debugreg6 & DR_TRAP_BITS) ||
+ (dr6 & (~DR_TRAP_BITS)))
rc = NOTIFY_DONE;
set_debugreg(dr7, 7);
--
1.6.2.3
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [GIT PULL] perf events updates
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
` (4 preceding siblings ...)
2010-01-29 2:07 ` [PATCH 5/5] x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler Frederic Weisbecker
@ 2010-01-29 7:06 ` Ingo Molnar
5 siblings, 0 replies; 8+ messages in thread
From: Ingo Molnar @ 2010-01-29 7:06 UTC (permalink / raw)
To: Frederic Weisbecker
Cc: LKML, K . Prasad, Xiao Guangrong, Paul Mackerras, Peter Zijlstra,
Anton Blanchard
* Frederic Weisbecker <fweisbec@gmail.com> wrote:
> Ingo,
>
> Please pull the perf/core branch that can be found at:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/frederic/random-tracing.git
> perf/core
>
> Thanks,
> Frederic
> ---
>
> K.Prasad (2):
> x86/debug: Clear reserved bits of DR6 in do_debug()
> x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler
>
> Xiao Guangrong (2):
> perf: Factorize trace events raw sample buffer operations
> tracing/kprobe: Cleanup unused return value of tracing functions
>
> Anton Blanchard (1):
> perf: Fix inconsistency between IP and callchain sampling
>
>
> arch/powerpc/kernel/perf_callchain.c | 3 -
> arch/sh/kernel/perf_callchain.c | 3 -
> arch/x86/include/asm/debugreg.h | 3 +
> arch/x86/kernel/cpu/perf_event.c | 3 -
> arch/x86/kernel/hw_breakpoint.c | 10 ++-
> arch/x86/kernel/traps.c | 3 +
> include/linux/ftrace_event.h | 18 +++++-
> include/trace/ftrace.h | 48 ++-------------
> kernel/trace/trace_event_profile.c | 52 +++++++++++++++--
> kernel/trace/trace_kprobe.c | 109 ++++++----------------------------
> kernel/trace/trace_syscalls.c | 71 +++-------------------
> 11 files changed, 109 insertions(+), 214 deletions(-)
Pulled, thanks a lot Frederic!
Ingo
^ permalink raw reply [flat|nested] 8+ messages in thread
* [GIT PULL] perf events updates
@ 2009-12-10 19:25 Ingo Molnar
0 siblings, 0 replies; 8+ messages in thread
From: Ingo Molnar @ 2009-12-10 19:25 UTC (permalink / raw)
To: Linus Torvalds
Cc: linux-kernel, Peter Zijlstra, Fr??d??ric Weisbecker,
Arnaldo Carvalho de Melo, Andrew Morton
Linus,
Please pull the latest perf events git tree from:
git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git perf-fixes-for-linus
Fixes mostly, plus some late changes/cleanups.
Thanks,
Ingo
------------------>
Arnaldo Carvalho de Melo (3):
perf buildid-list: Fix copy'n'paste help message
perf symbols: dsos__read_build_ids() should read both user and kernel buildids
perf symbols: perf_header__read_build_ids() offset'n'size should be u64
Arnd Bergmann (1):
perf events: hw_breakpoints: Don't include asm/hw_breakpoint.h in user space
Cyrill Gorcunov (1):
x86, perf events: Check if we have APIC enabled
Frederic Weisbecker (14):
hw-breakpoints: Add two reserved fields for future extensions
perf: Remove pointless union that wraps the hw breakpoint fields
perf: Remove unused struct perf_event::event_callback
hw-breakpoints: Drop callback and task parameters from modify helper
hw-breakpoints: Use overflow handler instead of the event callback
perf: Remove the "event" callback from perf events
x86/perf: Exclude the debug stack from the callchains
x86: Fixup wrong debug exception frame link in stacktraces
x86: Fixup wrong irq frame link in stacktraces
hw-breakpoints: Zeroe the breakpoint attrs on initialization
hw-breakpoints: Fix task-bound breakpoint slot allocation
hw-breakpoints: Modify breakpoints without unregistering them
tracing/kprobes: Fix field creation's bad error handling
perf sched: Add max delay time snapshot
Ingo Molnar (1):
perf kmem: Fix unused argument build warning
Jean Delvare (1):
x86, perf probe: Fix warning in test_get_len()
Juha Leppanen (1):
perf probe: Fix strtailcmp() to compare s1and s2[0]
Julia Lawall (2):
perf tools: Correct size computation in tracepoint_id_to_path()
perf tools: Correct size given to memset
Li Zefan (2):
perf kmem: Show usage if no option is specified
perf tools: Align long options which have no short forms
Masami Hiramatsu (12):
x86 insn: Delete empty or incomplete inat-tables.c
perf probe: Fix event namelist to duplicate string
perf probe: Check e_snprintf() format string
perf probe: Use pr_debug for debug message
perf probe: Change event list format
perf probe: Change probe-added message more user-friendly
perf probe: Fix add-probe command syntax without --add option
perf probe: Remove event suffix number _0
perf probe: Support vmlinux on cwd by default
trace-kprobe: Support delete probe syntax
perf probe: Support --del option
perf probe: Update perf-probe document
Mike Frysinger (1):
sched: Mark sched_clock() as notrace
OGAWA Hirofumi (4):
perf timechart: Fix header handling
perf: Make common SAMPLE_EVENT parser
perf tools: Misc small fixes
x86: Compile insn.c and inat.c only for KPROBES
Stephen Rothwell (1):
perf: hw_breakpoints: Fix percpu namespace clash
Tom Zanussi (2):
perf trace/scripting: Fix compile error when libperl not installed
perf trace/scripting: Don't display 'scripting unsupported' msg unnecessarily
Ulrich Drepper (1):
perf tools: Optimize parse_subsystem_tracepoint_event()
Xiao Guangrong (10):
perf/sched: Fix 'perf sched trace'
perf_event: Fix raw event processing
perf_event: Fix __dsos__write_buildid_table()
perf_event: Eliminate raw->size
perf_event: Clean up __perf_event_init_context()
perf_event: Allocate children's perf_event_ctxp at the right time
perf_event: Cleanup for cpu_clock_perf_event_update()
perf sched: Fix for getting task's execution time
perf_event: Fix perf_swevent_hrtimer() variable initialization
perf_event: Fix variable initialization in other codepaths
arch/x86/Kconfig.debug | 4 +-
arch/x86/kernel/cpu/perf_event.c | 31 +++--
arch/x86/kernel/dumpstack_64.c | 33 +++++-
arch/x86/kernel/entry_64.S | 6 +-
arch/x86/kernel/hw_breakpoint.c | 5 +-
arch/x86/kernel/ptrace.c | 68 +++++-----
arch/x86/lib/Makefile | 4 +-
arch/x86/tools/test_get_len.c | 2 +-
include/linux/hw_breakpoint.h | 40 +++----
include/linux/perf_event.h | 41 +++---
include/linux/sched.h | 3 +-
kernel/hw_breakpoint.c | 146 +++++++++++++---------
kernel/perf_event.c | 75 ++++++------
kernel/trace/trace_kprobe.c | 41 +++++--
kernel/trace/trace_ksym.c | 5 +-
samples/hw_breakpoint/data_breakpoint.c | 7 +-
tools/perf/Documentation/perf-kmem.txt | 13 +-
tools/perf/Documentation/perf-probe.txt | 21 +++-
tools/perf/builtin-buildid-list.c | 2 +-
tools/perf/builtin-kmem.c | 123 ++++++++----------
tools/perf/builtin-probe.c | 80 +++++++++---
tools/perf/builtin-report.c | 39 +++---
tools/perf/builtin-sched.c | 212 ++++++++++++++-----------------
tools/perf/builtin-timechart.c | 56 +++------
tools/perf/builtin-trace.c | 48 ++-----
tools/perf/util/data_map.c | 4 +-
tools/perf/util/data_map.h | 2 +-
tools/perf/util/event.c | 67 ++++++++++
tools/perf/util/event.h | 17 +++-
| 9 +-
tools/perf/util/parse-events.c | 17 +--
tools/perf/util/parse-options.c | 3 +
tools/perf/util/probe-event.c | 133 +++++++++++++++++--
tools/perf/util/probe-event.h | 1 +
tools/perf/util/probe-finder.c | 2 +-
tools/perf/util/symbol.c | 5 +-
tools/perf/util/trace-event-parse.c | 4 +-
tools/perf/util/trace-event-perl.c | 67 ++++++++--
tools/perf/util/trace-event-perl.h | 4 +
tools/perf/util/trace-event-read.c | 3 +-
40 files changed, 878 insertions(+), 565 deletions(-)
diff --git a/arch/x86/Kconfig.debug b/arch/x86/Kconfig.debug
index 7d0b681..0e90929 100644
--- a/arch/x86/Kconfig.debug
+++ b/arch/x86/Kconfig.debug
@@ -187,8 +187,8 @@ config HAVE_MMIOTRACE_SUPPORT
def_bool y
config X86_DECODER_SELFTEST
- bool "x86 instruction decoder selftest"
- depends on DEBUG_KERNEL
+ bool "x86 instruction decoder selftest"
+ depends on DEBUG_KERNEL && KPROBES
---help---
Perform x86 instruction decoder selftests at build time.
This option is useful for checking the sanity of x86 instruction
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index c1bbed1..18f05ec 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -1632,6 +1632,7 @@ static void intel_pmu_drain_bts_buffer(struct cpu_hw_events *cpuc)
data.period = event->hw.last_period;
data.addr = 0;
+ data.raw = NULL;
regs.ip = 0;
/*
@@ -1749,6 +1750,7 @@ static int p6_pmu_handle_irq(struct pt_regs *regs)
u64 val;
data.addr = 0;
+ data.raw = NULL;
cpuc = &__get_cpu_var(cpu_hw_events);
@@ -1794,6 +1796,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
u64 ack, status;
data.addr = 0;
+ data.raw = NULL;
cpuc = &__get_cpu_var(cpu_hw_events);
@@ -1857,6 +1860,7 @@ static int amd_pmu_handle_irq(struct pt_regs *regs)
u64 val;
data.addr = 0;
+ data.raw = NULL;
cpuc = &__get_cpu_var(cpu_hw_events);
@@ -2062,12 +2066,6 @@ static __init int p6_pmu_init(void)
x86_pmu = p6_pmu;
- if (!cpu_has_apic) {
- pr_info("no APIC, boot with the \"lapic\" boot parameter to force-enable it.\n");
- pr_info("no hardware sampling interrupt available.\n");
- x86_pmu.apic = 0;
- }
-
return 0;
}
@@ -2159,6 +2157,16 @@ static __init int amd_pmu_init(void)
return 0;
}
+static void __init pmu_check_apic(void)
+{
+ if (cpu_has_apic)
+ return;
+
+ x86_pmu.apic = 0;
+ pr_info("no APIC, boot with the \"lapic\" boot parameter to force-enable it.\n");
+ pr_info("no hardware sampling interrupt available.\n");
+}
+
void __init init_hw_perf_events(void)
{
int err;
@@ -2180,6 +2188,8 @@ void __init init_hw_perf_events(void)
return;
}
+ pmu_check_apic();
+
pr_cont("%s PMU driver.\n", x86_pmu.name);
if (x86_pmu.num_events > X86_PMC_MAX_GENERIC) {
@@ -2287,7 +2297,7 @@ void callchain_store(struct perf_callchain_entry *entry, u64 ip)
static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_irq_entry);
static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_nmi_entry);
-static DEFINE_PER_CPU(int, in_nmi_frame);
+static DEFINE_PER_CPU(int, in_ignored_frame);
static void
@@ -2303,8 +2313,9 @@ static void backtrace_warning(void *data, char *msg)
static int backtrace_stack(void *data, char *name)
{
- per_cpu(in_nmi_frame, smp_processor_id()) =
- x86_is_stack_id(NMI_STACK, name);
+ per_cpu(in_ignored_frame, smp_processor_id()) =
+ x86_is_stack_id(NMI_STACK, name) ||
+ x86_is_stack_id(DEBUG_STACK, name);
return 0;
}
@@ -2313,7 +2324,7 @@ static void backtrace_address(void *data, unsigned long addr, int reliable)
{
struct perf_callchain_entry *entry = data;
- if (per_cpu(in_nmi_frame, smp_processor_id()))
+ if (per_cpu(in_ignored_frame, smp_processor_id()))
return;
if (reliable)
diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c
index a071e6b..004b8aa 100644
--- a/arch/x86/kernel/dumpstack_64.c
+++ b/arch/x86/kernel/dumpstack_64.c
@@ -101,6 +101,35 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
return NULL;
}
+static inline int
+in_irq_stack(unsigned long *stack, unsigned long *irq_stack,
+ unsigned long *irq_stack_end)
+{
+ return (stack >= irq_stack && stack < irq_stack_end);
+}
+
+/*
+ * We are returning from the irq stack and go to the previous one.
+ * If the previous stack is also in the irq stack, then bp in the first
+ * frame of the irq stack points to the previous, interrupted one.
+ * Otherwise we have another level of indirection: We first save
+ * the bp of the previous stack, then we switch the stack to the irq one
+ * and save a new bp that links to the previous one.
+ * (See save_args())
+ */
+static inline unsigned long
+fixup_bp_irq_link(unsigned long bp, unsigned long *stack,
+ unsigned long *irq_stack, unsigned long *irq_stack_end)
+{
+#ifdef CONFIG_FRAME_POINTER
+ struct stack_frame *frame = (struct stack_frame *)bp;
+
+ if (!in_irq_stack(stack, irq_stack, irq_stack_end))
+ return (unsigned long)frame->next_frame;
+#endif
+ return bp;
+}
+
/*
* x86-64 can have up to three kernel stacks:
* process stack
@@ -173,7 +202,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
irq_stack = irq_stack_end -
(IRQ_STACK_SIZE - 64) / sizeof(*irq_stack);
- if (stack >= irq_stack && stack < irq_stack_end) {
+ if (in_irq_stack(stack, irq_stack, irq_stack_end)) {
if (ops->stack(data, "IRQ") < 0)
break;
bp = print_context_stack(tinfo, stack, bp,
@@ -184,6 +213,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs,
* pointer (index -1 to end) in the IRQ stack:
*/
stack = (unsigned long *) (irq_stack_end[-1]);
+ bp = fixup_bp_irq_link(bp, stack, irq_stack,
+ irq_stack_end);
irq_stack_end = NULL;
ops->stack(data, "EOI");
continue;
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index 722df1b..0f08a0c 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -1076,10 +1076,10 @@ ENTRY(\sym)
TRACE_IRQS_OFF
movq %rsp,%rdi /* pt_regs pointer */
xorl %esi,%esi /* no error code */
- PER_CPU(init_tss, %rbp)
- subq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%rbp)
+ PER_CPU(init_tss, %r12)
+ subq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%r12)
call \do_sym
- addq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%rbp)
+ addq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%r12)
jmp paranoid_exit /* %ebx: no swapgs flag */
CFI_ENDPROC
END(\sym)
diff --git a/arch/x86/kernel/hw_breakpoint.c b/arch/x86/kernel/hw_breakpoint.c
index d42f65a..05d5fec 100644
--- a/arch/x86/kernel/hw_breakpoint.c
+++ b/arch/x86/kernel/hw_breakpoint.c
@@ -362,8 +362,7 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp,
return ret;
}
- if (bp->callback)
- ret = arch_store_info(bp);
+ ret = arch_store_info(bp);
if (ret < 0)
return ret;
@@ -519,7 +518,7 @@ static int __kprobes hw_breakpoint_handler(struct die_args *args)
break;
}
- (bp->callback)(bp, args->regs);
+ perf_bp_event(bp, args->regs);
rcu_read_unlock();
}
diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c
index 04d182a..7079dda 100644
--- a/arch/x86/kernel/ptrace.c
+++ b/arch/x86/kernel/ptrace.c
@@ -555,7 +555,9 @@ static int genregs_set(struct task_struct *target,
return ret;
}
-static void ptrace_triggered(struct perf_event *bp, void *data)
+static void ptrace_triggered(struct perf_event *bp, int nmi,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
{
int i;
struct thread_struct *thread = &(current->thread);
@@ -593,13 +595,13 @@ static unsigned long ptrace_get_dr7(struct perf_event *bp[])
return dr7;
}
-static struct perf_event *
+static int
ptrace_modify_breakpoint(struct perf_event *bp, int len, int type,
struct task_struct *tsk, int disabled)
{
int err;
int gen_len, gen_type;
- DEFINE_BREAKPOINT_ATTR(attr);
+ struct perf_event_attr attr;
/*
* We shoud have at least an inactive breakpoint at this
@@ -607,18 +609,18 @@ ptrace_modify_breakpoint(struct perf_event *bp, int len, int type,
* written the address register first
*/
if (!bp)
- return ERR_PTR(-EINVAL);
+ return -EINVAL;
err = arch_bp_generic_fields(len, type, &gen_len, &gen_type);
if (err)
- return ERR_PTR(err);
+ return err;
attr = bp->attr;
attr.bp_len = gen_len;
attr.bp_type = gen_type;
attr.disabled = disabled;
- return modify_user_hw_breakpoint(bp, &attr, bp->callback, tsk);
+ return modify_user_hw_breakpoint(bp, &attr);
}
/*
@@ -656,28 +658,17 @@ restore:
if (!second_pass)
continue;
- thread->ptrace_bps[i] = NULL;
- bp = ptrace_modify_breakpoint(bp, len, type,
+ rc = ptrace_modify_breakpoint(bp, len, type,
tsk, 1);
- if (IS_ERR(bp)) {
- rc = PTR_ERR(bp);
- thread->ptrace_bps[i] = NULL;
+ if (rc)
break;
- }
- thread->ptrace_bps[i] = bp;
}
continue;
}
- bp = ptrace_modify_breakpoint(bp, len, type, tsk, 0);
-
- /* Incorrect bp, or we have a bug in bp API */
- if (IS_ERR(bp)) {
- rc = PTR_ERR(bp);
- thread->ptrace_bps[i] = NULL;
+ rc = ptrace_modify_breakpoint(bp, len, type, tsk, 0);
+ if (rc)
break;
- }
- thread->ptrace_bps[i] = bp;
}
/*
* Make a second pass to free the remaining unused breakpoints
@@ -721,9 +712,10 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr,
{
struct perf_event *bp;
struct thread_struct *t = &tsk->thread;
- DEFINE_BREAKPOINT_ATTR(attr);
+ struct perf_event_attr attr;
if (!t->ptrace_bps[nr]) {
+ hw_breakpoint_init(&attr);
/*
* Put stub len and type to register (reserve) an inactive but
* correct bp
@@ -734,26 +726,32 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr,
attr.disabled = 1;
bp = register_user_hw_breakpoint(&attr, ptrace_triggered, tsk);
+
+ /*
+ * CHECKME: the previous code returned -EIO if the addr wasn't
+ * a valid task virtual addr. The new one will return -EINVAL in
+ * this case.
+ * -EINVAL may be what we want for in-kernel breakpoints users,
+ * but -EIO looks better for ptrace, since we refuse a register
+ * writing for the user. And anyway this is the previous
+ * behaviour.
+ */
+ if (IS_ERR(bp))
+ return PTR_ERR(bp);
+
+ t->ptrace_bps[nr] = bp;
} else {
+ int err;
+
bp = t->ptrace_bps[nr];
- t->ptrace_bps[nr] = NULL;
attr = bp->attr;
attr.bp_addr = addr;
- bp = modify_user_hw_breakpoint(bp, &attr, bp->callback, tsk);
+ err = modify_user_hw_breakpoint(bp, &attr);
+ if (err)
+ return err;
}
- /*
- * CHECKME: the previous code returned -EIO if the addr wasn't a
- * valid task virtual addr. The new one will return -EINVAL in this
- * case.
- * -EINVAL may be what we want for in-kernel breakpoints users, but
- * -EIO looks better for ptrace, since we refuse a register writing
- * for the user. And anyway this is the previous behaviour.
- */
- if (IS_ERR(bp))
- return PTR_ERR(bp);
- t->ptrace_bps[nr] = bp;
return 0;
}
diff --git a/arch/x86/lib/Makefile b/arch/x86/lib/Makefile
index a2d6472..45b20e4 100644
--- a/arch/x86/lib/Makefile
+++ b/arch/x86/lib/Makefile
@@ -5,7 +5,7 @@
inat_tables_script = $(srctree)/arch/x86/tools/gen-insn-attr-x86.awk
inat_tables_maps = $(srctree)/arch/x86/lib/x86-opcode-map.txt
quiet_cmd_inat_tables = GEN $@
- cmd_inat_tables = $(AWK) -f $(inat_tables_script) $(inat_tables_maps) > $@
+ cmd_inat_tables = $(AWK) -f $(inat_tables_script) $(inat_tables_maps) > $@ || rm -f $@
$(obj)/inat-tables.c: $(inat_tables_script) $(inat_tables_maps)
$(call cmd,inat_tables)
@@ -20,7 +20,7 @@ lib-y := delay.o
lib-y += thunk_$(BITS).o
lib-y += usercopy_$(BITS).o getuser.o putuser.o
lib-y += memcpy_$(BITS).o
-lib-y += insn.o inat.o
+lib-$(CONFIG_KPROBES) += insn.o inat.o
obj-y += msr-reg.o msr-reg-export.o
diff --git a/arch/x86/tools/test_get_len.c b/arch/x86/tools/test_get_len.c
index d8214dc..bee8d6a 100644
--- a/arch/x86/tools/test_get_len.c
+++ b/arch/x86/tools/test_get_len.c
@@ -113,7 +113,7 @@ int main(int argc, char **argv)
char line[BUFSIZE], sym[BUFSIZE] = "<unknown>";
unsigned char insn_buf[16];
struct insn insn;
- int insns = 0, c;
+ int insns = 0;
int warnings = 0;
parse_args(argc, argv);
diff --git a/include/linux/hw_breakpoint.h b/include/linux/hw_breakpoint.h
index a03daed..69f07a9 100644
--- a/include/linux/hw_breakpoint.h
+++ b/include/linux/hw_breakpoint.h
@@ -20,19 +20,18 @@ enum {
#ifdef CONFIG_HAVE_HW_BREAKPOINT
-/* As it's for in-kernel or ptrace use, we want it to be pinned */
-#define DEFINE_BREAKPOINT_ATTR(name) \
-struct perf_event_attr name = { \
- .type = PERF_TYPE_BREAKPOINT, \
- .size = sizeof(name), \
- .pinned = 1, \
-};
-
static inline void hw_breakpoint_init(struct perf_event_attr *attr)
{
+ memset(attr, 0, sizeof(*attr));
+
attr->type = PERF_TYPE_BREAKPOINT;
attr->size = sizeof(*attr);
+ /*
+ * As it's for in-kernel or ptrace use, we want it to be pinned
+ * and to call its callback every hits.
+ */
attr->pinned = 1;
+ attr->sample_period = 1;
}
static inline unsigned long hw_breakpoint_addr(struct perf_event *bp)
@@ -52,27 +51,24 @@ static inline int hw_breakpoint_len(struct perf_event *bp)
extern struct perf_event *
register_user_hw_breakpoint(struct perf_event_attr *attr,
- perf_callback_t triggered,
+ perf_overflow_handler_t triggered,
struct task_struct *tsk);
/* FIXME: only change from the attr, and don't unregister */
-extern struct perf_event *
-modify_user_hw_breakpoint(struct perf_event *bp,
- struct perf_event_attr *attr,
- perf_callback_t triggered,
- struct task_struct *tsk);
+extern int
+modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *attr);
/*
* Kernel breakpoints are not associated with any particular thread.
*/
extern struct perf_event *
register_wide_hw_breakpoint_cpu(struct perf_event_attr *attr,
- perf_callback_t triggered,
+ perf_overflow_handler_t triggered,
int cpu);
extern struct perf_event **
register_wide_hw_breakpoint(struct perf_event_attr *attr,
- perf_callback_t triggered);
+ perf_overflow_handler_t triggered);
extern int register_perf_hw_breakpoint(struct perf_event *bp);
extern int __register_perf_hw_breakpoint(struct perf_event *bp);
@@ -93,20 +89,18 @@ static inline struct arch_hw_breakpoint *counter_arch_bp(struct perf_event *bp)
static inline struct perf_event *
register_user_hw_breakpoint(struct perf_event_attr *attr,
- perf_callback_t triggered,
+ perf_overflow_handler_t triggered,
struct task_struct *tsk) { return NULL; }
-static inline struct perf_event *
+static inline int
modify_user_hw_breakpoint(struct perf_event *bp,
- struct perf_event_attr *attr,
- perf_callback_t triggered,
- struct task_struct *tsk) { return NULL; }
+ struct perf_event_attr *attr) { return NULL; }
static inline struct perf_event *
register_wide_hw_breakpoint_cpu(struct perf_event_attr *attr,
- perf_callback_t triggered,
+ perf_overflow_handler_t triggered,
int cpu) { return NULL; }
static inline struct perf_event **
register_wide_hw_breakpoint(struct perf_event_attr *attr,
- perf_callback_t triggered) { return NULL; }
+ perf_overflow_handler_t triggered) { return NULL; }
static inline int
register_perf_hw_breakpoint(struct perf_event *bp) { return -ENOSYS; }
static inline int
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 43adbd7..64a53f7 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -18,10 +18,6 @@
#include <linux/ioctl.h>
#include <asm/byteorder.h>
-#ifdef CONFIG_HAVE_HW_BREAKPOINT
-#include <asm/hw_breakpoint.h>
-#endif
-
/*
* User-space ABI bits:
*/
@@ -215,12 +211,12 @@ struct perf_event_attr {
__u32 wakeup_watermark; /* bytes before wakeup */
};
- union {
- struct { /* Hardware breakpoint info */
- __u64 bp_addr;
- __u32 bp_type;
- __u32 bp_len;
- };
+ struct { /* Hardware breakpoint info */
+ __u64 bp_addr;
+ __u32 bp_type;
+ __u32 bp_len;
+ __u64 __bp_reserved_1;
+ __u64 __bp_reserved_2;
};
__u32 __reserved_2;
@@ -451,6 +447,10 @@ enum perf_callchain_context {
# include <asm/perf_event.h>
#endif
+#ifdef CONFIG_HAVE_HW_BREAKPOINT
+#include <asm/hw_breakpoint.h>
+#endif
+
#include <linux/list.h>
#include <linux/mutex.h>
#include <linux/rculist.h>
@@ -565,10 +565,12 @@ struct perf_pending_entry {
void (*func)(struct perf_pending_entry *);
};
-typedef void (*perf_callback_t)(struct perf_event *, void *);
-
struct perf_sample_data;
+typedef void (*perf_overflow_handler_t)(struct perf_event *, int,
+ struct perf_sample_data *,
+ struct pt_regs *regs);
+
/**
* struct perf_event - performance event kernel representation:
*/
@@ -660,18 +662,12 @@ struct perf_event {
struct pid_namespace *ns;
u64 id;
- void (*overflow_handler)(struct perf_event *event,
- int nmi, struct perf_sample_data *data,
- struct pt_regs *regs);
+ perf_overflow_handler_t overflow_handler;
#ifdef CONFIG_EVENT_PROFILE
struct event_filter *filter;
#endif
- perf_callback_t callback;
-
- perf_callback_t event_callback;
-
#endif /* CONFIG_PERF_EVENTS */
};
@@ -781,7 +777,7 @@ extern struct perf_event *
perf_event_create_kernel_counter(struct perf_event_attr *attr,
int cpu,
pid_t pid,
- perf_callback_t callback);
+ perf_overflow_handler_t callback);
extern u64 perf_event_read_value(struct perf_event *event,
u64 *enabled, u64 *running);
@@ -876,6 +872,8 @@ extern void perf_output_copy(struct perf_output_handle *handle,
const void *buf, unsigned int len);
extern int perf_swevent_get_recursion_context(void);
extern void perf_swevent_put_recursion_context(int rctx);
+extern void perf_event_enable(struct perf_event *event);
+extern void perf_event_disable(struct perf_event *event);
#else
static inline void
perf_event_task_sched_in(struct task_struct *task, int cpu) { }
@@ -906,7 +904,8 @@ static inline void perf_event_fork(struct task_struct *tsk) { }
static inline void perf_event_init(void) { }
static inline int perf_swevent_get_recursion_context(void) { return -1; }
static inline void perf_swevent_put_recursion_context(int rctx) { }
-
+static inline void perf_event_enable(struct perf_event *event) { }
+static inline void perf_event_disable(struct perf_event *event) { }
#endif
#define perf_output_put(handle, x) \
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 75e6e60..576d838 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1836,7 +1836,8 @@ static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
extern int sched_clock_stable;
#endif
-extern unsigned long long sched_clock(void);
+/* ftrace calls sched_clock() directly */
+extern unsigned long long notrace sched_clock(void);
extern void sched_clock_init(void);
extern u64 sched_clock_cpu(int cpu);
diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c
index cf5ee16..366eedf 100644
--- a/kernel/hw_breakpoint.c
+++ b/kernel/hw_breakpoint.c
@@ -52,7 +52,7 @@
static DEFINE_PER_CPU(unsigned int, nr_cpu_bp_pinned);
/* Number of pinned task breakpoints in a cpu */
-static DEFINE_PER_CPU(unsigned int, task_bp_pinned[HBP_NUM]);
+static DEFINE_PER_CPU(unsigned int, nr_task_bp_pinned[HBP_NUM]);
/* Number of non-pinned cpu/task breakpoints in a cpu */
static DEFINE_PER_CPU(unsigned int, nr_bp_flexible);
@@ -73,7 +73,7 @@ static DEFINE_MUTEX(nr_bp_mutex);
static unsigned int max_task_bp_pinned(int cpu)
{
int i;
- unsigned int *tsk_pinned = per_cpu(task_bp_pinned, cpu);
+ unsigned int *tsk_pinned = per_cpu(nr_task_bp_pinned, cpu);
for (i = HBP_NUM -1; i >= 0; i--) {
if (tsk_pinned[i] > 0)
@@ -83,15 +83,51 @@ static unsigned int max_task_bp_pinned(int cpu)
return 0;
}
+static int task_bp_pinned(struct task_struct *tsk)
+{
+ struct perf_event_context *ctx = tsk->perf_event_ctxp;
+ struct list_head *list;
+ struct perf_event *bp;
+ unsigned long flags;
+ int count = 0;
+
+ if (WARN_ONCE(!ctx, "No perf context for this task"))
+ return 0;
+
+ list = &ctx->event_list;
+
+ spin_lock_irqsave(&ctx->lock, flags);
+
+ /*
+ * The current breakpoint counter is not included in the list
+ * at the open() callback time
+ */
+ list_for_each_entry(bp, list, event_entry) {
+ if (bp->attr.type == PERF_TYPE_BREAKPOINT)
+ count++;
+ }
+
+ spin_unlock_irqrestore(&ctx->lock, flags);
+
+ return count;
+}
+
/*
* Report the number of pinned/un-pinned breakpoints we have in
* a given cpu (cpu > -1) or in all of them (cpu = -1).
*/
-static void fetch_bp_busy_slots(struct bp_busy_slots *slots, int cpu)
+static void
+fetch_bp_busy_slots(struct bp_busy_slots *slots, struct perf_event *bp)
{
+ int cpu = bp->cpu;
+ struct task_struct *tsk = bp->ctx->task;
+
if (cpu >= 0) {
slots->pinned = per_cpu(nr_cpu_bp_pinned, cpu);
- slots->pinned += max_task_bp_pinned(cpu);
+ if (!tsk)
+ slots->pinned += max_task_bp_pinned(cpu);
+ else
+ slots->pinned += task_bp_pinned(tsk);
slots->flexible = per_cpu(nr_bp_flexible, cpu);
return;
@@ -101,7 +137,10 @@ static void fetch_bp_busy_slots(struct bp_busy_slots *slots, int cpu)
unsigned int nr;
nr = per_cpu(nr_cpu_bp_pinned, cpu);
- nr += max_task_bp_pinned(cpu);
+ if (!tsk)
+ nr += max_task_bp_pinned(cpu);
+ else
+ nr += task_bp_pinned(tsk);
if (nr > slots->pinned)
slots->pinned = nr;
@@ -118,35 +157,12 @@ static void fetch_bp_busy_slots(struct bp_busy_slots *slots, int cpu)
*/
static void toggle_bp_task_slot(struct task_struct *tsk, int cpu, bool enable)
{
- int count = 0;
- struct perf_event *bp;
- struct perf_event_context *ctx = tsk->perf_event_ctxp;
unsigned int *tsk_pinned;
- struct list_head *list;
- unsigned long flags;
-
- if (WARN_ONCE(!ctx, "No perf context for this task"))
- return;
-
- list = &ctx->event_list;
-
- spin_lock_irqsave(&ctx->lock, flags);
-
- /*
- * The current breakpoint counter is not included in the list
- * at the open() callback time
- */
- list_for_each_entry(bp, list, event_entry) {
- if (bp->attr.type == PERF_TYPE_BREAKPOINT)
- count++;
- }
+ int count = 0;
- spin_unlock_irqrestore(&ctx->lock, flags);
+ count = task_bp_pinned(tsk);
- if (WARN_ONCE(count < 0, "No breakpoint counter found in the counter list"))
- return;
-
- tsk_pinned = per_cpu(task_bp_pinned, cpu);
+ tsk_pinned = per_cpu(nr_task_bp_pinned, cpu);
if (enable) {
tsk_pinned[count]++;
if (count > 0)
@@ -193,7 +209,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
* - If attached to a single cpu, check:
*
* (per_cpu(nr_bp_flexible, cpu) || (per_cpu(nr_cpu_bp_pinned, cpu)
- * + max(per_cpu(task_bp_pinned, cpu)))) < HBP_NUM
+ * + max(per_cpu(nr_task_bp_pinned, cpu)))) < HBP_NUM
*
* -> If there are already non-pinned counters in this cpu, it means
* there is already a free slot for them.
@@ -204,7 +220,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
* - If attached to every cpus, check:
*
* (per_cpu(nr_bp_flexible, *) || (max(per_cpu(nr_cpu_bp_pinned, *))
- * + max(per_cpu(task_bp_pinned, *)))) < HBP_NUM
+ * + max(per_cpu(nr_task_bp_pinned, *)))) < HBP_NUM
*
* -> This is roughly the same, except we check the number of per cpu
* bp for every cpu and we keep the max one. Same for the per tasks
@@ -216,7 +232,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
* - If attached to a single cpu, check:
*
* ((per_cpu(nr_bp_flexible, cpu) > 1) + per_cpu(nr_cpu_bp_pinned, cpu)
- * + max(per_cpu(task_bp_pinned, cpu))) < HBP_NUM
+ * + max(per_cpu(nr_task_bp_pinned, cpu))) < HBP_NUM
*
* -> Same checks as before. But now the nr_bp_flexible, if any, must keep
* one register at least (or they will never be fed).
@@ -224,7 +240,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
* - If attached to every cpus, check:
*
* ((per_cpu(nr_bp_flexible, *) > 1) + max(per_cpu(nr_cpu_bp_pinned, *))
- * + max(per_cpu(task_bp_pinned, *))) < HBP_NUM
+ * + max(per_cpu(nr_task_bp_pinned, *))) < HBP_NUM
*/
int reserve_bp_slot(struct perf_event *bp)
{
@@ -233,7 +249,7 @@ int reserve_bp_slot(struct perf_event *bp)
mutex_lock(&nr_bp_mutex);
- fetch_bp_busy_slots(&slots, bp->cpu);
+ fetch_bp_busy_slots(&slots, bp);
/* Flexible counters need to keep at least one slot */
if (slots.pinned + (!!slots.flexible) == HBP_NUM) {
@@ -259,7 +275,7 @@ void release_bp_slot(struct perf_event *bp)
}
-int __register_perf_hw_breakpoint(struct perf_event *bp)
+int register_perf_hw_breakpoint(struct perf_event *bp)
{
int ret;
@@ -276,19 +292,12 @@ int __register_perf_hw_breakpoint(struct perf_event *bp)
* This is a quick hack that will be removed soon, once we remove
* the tmp breakpoints from ptrace
*/
- if (!bp->attr.disabled || bp->callback == perf_bp_event)
+ if (!bp->attr.disabled || !bp->overflow_handler)
ret = arch_validate_hwbkpt_settings(bp, bp->ctx->task);
return ret;
}
-int register_perf_hw_breakpoint(struct perf_event *bp)
-{
- bp->callback = perf_bp_event;
-
- return __register_perf_hw_breakpoint(bp);
-}
-
/**
* register_user_hw_breakpoint - register a hardware breakpoint for user space
* @attr: breakpoint attributes
@@ -297,7 +306,7 @@ int register_perf_hw_breakpoint(struct perf_event *bp)
*/
struct perf_event *
register_user_hw_breakpoint(struct perf_event_attr *attr,
- perf_callback_t triggered,
+ perf_overflow_handler_t triggered,
struct task_struct *tsk)
{
return perf_event_create_kernel_counter(attr, -1, tsk->pid, triggered);
@@ -311,19 +320,40 @@ EXPORT_SYMBOL_GPL(register_user_hw_breakpoint);
* @triggered: callback to trigger when we hit the breakpoint
* @tsk: pointer to 'task_struct' of the process to which the address belongs
*/
-struct perf_event *
-modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *attr,
- perf_callback_t triggered,
- struct task_struct *tsk)
+int modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *attr)
{
- /*
- * FIXME: do it without unregistering
- * - We don't want to lose our slot
- * - If the new bp is incorrect, don't lose the older one
- */
- unregister_hw_breakpoint(bp);
+ u64 old_addr = bp->attr.bp_addr;
+ int old_type = bp->attr.bp_type;
+ int old_len = bp->attr.bp_len;
+ int err = 0;
- return perf_event_create_kernel_counter(attr, -1, tsk->pid, triggered);
+ perf_event_disable(bp);
+
+ bp->attr.bp_addr = attr->bp_addr;
+ bp->attr.bp_type = attr->bp_type;
+ bp->attr.bp_len = attr->bp_len;
+
+ if (attr->disabled)
+ goto end;
+
+ err = arch_validate_hwbkpt_settings(bp, bp->ctx->task);
+ if (!err)
+ perf_event_enable(bp);
+
+ if (err) {
+ bp->attr.bp_addr = old_addr;
+ bp->attr.bp_type = old_type;
+ bp->attr.bp_len = old_len;
+ if (!bp->attr.disabled)
+ perf_event_enable(bp);
+
+ return err;
+ }
+
+end:
+ bp->attr.disabled = attr->disabled;
+
+ return 0;
}
EXPORT_SYMBOL_GPL(modify_user_hw_breakpoint);
@@ -348,7 +378,7 @@ EXPORT_SYMBOL_GPL(unregister_hw_breakpoint);
*/
struct perf_event **
register_wide_hw_breakpoint(struct perf_event_attr *attr,
- perf_callback_t triggered)
+ perf_overflow_handler_t triggered)
{
struct perf_event **cpu_events, **pevent, *bp;
long err;
diff --git a/kernel/perf_event.c b/kernel/perf_event.c
index 6b7ddba..d891ec4 100644
--- a/kernel/perf_event.c
+++ b/kernel/perf_event.c
@@ -36,7 +36,7 @@
/*
* Each CPU has a list of per CPU events:
*/
-DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context);
+static DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context);
int perf_max_events __read_mostly = 1;
static int perf_reserved_percpu __read_mostly;
@@ -567,7 +567,7 @@ static void __perf_event_disable(void *info)
* is the current context on this CPU and preemption is disabled,
* hence we can't get into perf_event_task_sched_out for this context.
*/
-static void perf_event_disable(struct perf_event *event)
+void perf_event_disable(struct perf_event *event)
{
struct perf_event_context *ctx = event->ctx;
struct task_struct *task = ctx->task;
@@ -971,7 +971,7 @@ static void __perf_event_enable(void *info)
* perf_event_for_each_child or perf_event_for_each as described
* for perf_event_disable.
*/
-static void perf_event_enable(struct perf_event *event)
+void perf_event_enable(struct perf_event *event)
{
struct perf_event_context *ctx = event->ctx;
struct task_struct *task = ctx->task;
@@ -1579,7 +1579,6 @@ static void
__perf_event_init_context(struct perf_event_context *ctx,
struct task_struct *task)
{
- memset(ctx, 0, sizeof(*ctx));
spin_lock_init(&ctx->lock);
mutex_init(&ctx->mutex);
INIT_LIST_HEAD(&ctx->group_list);
@@ -1654,7 +1653,7 @@ static struct perf_event_context *find_get_context(pid_t pid, int cpu)
}
if (!ctx) {
- ctx = kmalloc(sizeof(struct perf_event_context), GFP_KERNEL);
+ ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL);
err = -ENOMEM;
if (!ctx)
goto errout;
@@ -4011,6 +4010,7 @@ static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer)
event->pmu->read(event);
data.addr = 0;
+ data.raw = NULL;
data.period = event->hw.last_period;
regs = get_irq_regs();
/*
@@ -4080,8 +4080,7 @@ static void cpu_clock_perf_event_update(struct perf_event *event)
u64 now;
now = cpu_clock(cpu);
- prev = atomic64_read(&event->hw.prev_count);
- atomic64_set(&event->hw.prev_count, now);
+ prev = atomic64_xchg(&event->hw.prev_count, now);
atomic64_add(now - prev, &event->count);
}
@@ -4286,15 +4285,8 @@ static void bp_perf_event_destroy(struct perf_event *event)
static const struct pmu *bp_perf_event_init(struct perf_event *bp)
{
int err;
- /*
- * The breakpoint is already filled if we haven't created the counter
- * through perf syscall
- * FIXME: manage to get trigerred to NULL if it comes from syscalls
- */
- if (!bp->callback)
- err = register_perf_hw_breakpoint(bp);
- else
- err = __register_perf_hw_breakpoint(bp);
+
+ err = register_perf_hw_breakpoint(bp);
if (err)
return ERR_PTR(err);
@@ -4308,6 +4300,7 @@ void perf_bp_event(struct perf_event *bp, void *data)
struct perf_sample_data sample;
struct pt_regs *regs = data;
+ sample.raw = NULL;
sample.addr = bp->attr.bp_addr;
if (!perf_exclude_event(bp, regs))
@@ -4390,7 +4383,7 @@ perf_event_alloc(struct perf_event_attr *attr,
struct perf_event_context *ctx,
struct perf_event *group_leader,
struct perf_event *parent_event,
- perf_callback_t callback,
+ perf_overflow_handler_t overflow_handler,
gfp_t gfpflags)
{
const struct pmu *pmu;
@@ -4433,10 +4426,10 @@ perf_event_alloc(struct perf_event_attr *attr,
event->state = PERF_EVENT_STATE_INACTIVE;
- if (!callback && parent_event)
- callback = parent_event->callback;
+ if (!overflow_handler && parent_event)
+ overflow_handler = parent_event->overflow_handler;
- event->callback = callback;
+ event->overflow_handler = overflow_handler;
if (attr->disabled)
event->state = PERF_EVENT_STATE_OFF;
@@ -4776,7 +4769,8 @@ err_put_context:
*/
struct perf_event *
perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
- pid_t pid, perf_callback_t callback)
+ pid_t pid,
+ perf_overflow_handler_t overflow_handler)
{
struct perf_event *event;
struct perf_event_context *ctx;
@@ -4793,7 +4787,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
}
event = perf_event_alloc(attr, cpu, ctx, NULL,
- NULL, callback, GFP_KERNEL);
+ NULL, overflow_handler, GFP_KERNEL);
if (IS_ERR(event)) {
err = PTR_ERR(event);
goto err_put_context;
@@ -5090,7 +5084,7 @@ again:
*/
int perf_event_init_task(struct task_struct *child)
{
- struct perf_event_context *child_ctx, *parent_ctx;
+ struct perf_event_context *child_ctx = NULL, *parent_ctx;
struct perf_event_context *cloned_ctx;
struct perf_event *event;
struct task_struct *parent = current;
@@ -5106,20 +5100,6 @@ int perf_event_init_task(struct task_struct *child)
return 0;
/*
- * This is executed from the parent task context, so inherit
- * events that have been marked for cloning.
- * First allocate and initialize a context for the child.
- */
-
- child_ctx = kmalloc(sizeof(struct perf_event_context), GFP_KERNEL);
- if (!child_ctx)
- return -ENOMEM;
-
- __perf_event_init_context(child_ctx, child);
- child->perf_event_ctxp = child_ctx;
- get_task_struct(child);
-
- /*
* If the parent's context is a clone, pin it so it won't get
* swapped under us.
*/
@@ -5149,6 +5129,26 @@ int perf_event_init_task(struct task_struct *child)
continue;
}
+ if (!child->perf_event_ctxp) {
+ /*
+ * This is executed from the parent task context, so
+ * inherit events that have been marked for cloning.
+ * First allocate and initialize a context for the
+ * child.
+ */
+
+ child_ctx = kzalloc(sizeof(struct perf_event_context),
+ GFP_KERNEL);
+ if (!child_ctx) {
+ ret = -ENOMEM;
+ goto exit;
+ }
+
+ __perf_event_init_context(child_ctx, child);
+ child->perf_event_ctxp = child_ctx;
+ get_task_struct(child);
+ }
+
ret = inherit_group(event, parent, parent_ctx,
child, child_ctx);
if (ret) {
@@ -5177,6 +5177,7 @@ int perf_event_init_task(struct task_struct *child)
get_ctx(child_ctx->parent_ctx);
}
+exit:
mutex_unlock(&parent_ctx->mutex);
perf_unpin_context(parent_ctx);
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index aff5f80..b52d397 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -606,23 +606,22 @@ static int create_trace_probe(int argc, char **argv)
*/
struct trace_probe *tp;
int i, ret = 0;
- int is_return = 0;
+ int is_return = 0, is_delete = 0;
char *symbol = NULL, *event = NULL, *arg = NULL, *group = NULL;
unsigned long offset = 0;
void *addr = NULL;
char buf[MAX_EVENT_NAME_LEN];
- if (argc < 2) {
- pr_info("Probe point is not specified.\n");
- return -EINVAL;
- }
-
+ /* argc must be >= 1 */
if (argv[0][0] == 'p')
is_return = 0;
else if (argv[0][0] == 'r')
is_return = 1;
+ else if (argv[0][0] == '-')
+ is_delete = 1;
else {
- pr_info("Probe definition must be started with 'p' or 'r'.\n");
+ pr_info("Probe definition must be started with 'p', 'r' or"
+ " '-'.\n");
return -EINVAL;
}
@@ -642,7 +641,29 @@ static int create_trace_probe(int argc, char **argv)
return -EINVAL;
}
}
+ if (!group)
+ group = KPROBE_EVENT_SYSTEM;
+ if (is_delete) {
+ if (!event) {
+ pr_info("Delete command needs an event name.\n");
+ return -EINVAL;
+ }
+ tp = find_probe_event(event, group);
+ if (!tp) {
+ pr_info("Event %s/%s doesn't exist.\n", group, event);
+ return -ENOENT;
+ }
+ /* delete an event */
+ unregister_trace_probe(tp);
+ free_trace_probe(tp);
+ return 0;
+ }
+
+ if (argc < 2) {
+ pr_info("Probe point is not specified.\n");
+ return -EINVAL;
+ }
if (isdigit(argv[1][0])) {
if (is_return) {
pr_info("Return probe point must be a symbol.\n");
@@ -671,8 +692,6 @@ static int create_trace_probe(int argc, char **argv)
argc -= 2; argv += 2;
/* setup a probe */
- if (!group)
- group = KPROBE_EVENT_SYSTEM;
if (!event) {
/* Make a new event name */
if (symbol)
@@ -1114,7 +1133,7 @@ static int kprobe_event_define_fields(struct ftrace_event_call *event_call)
struct trace_probe *tp = (struct trace_probe *)event_call->data;
ret = trace_define_common_fields(event_call);
- if (!ret)
+ if (ret)
return ret;
DEFINE_FIELD(unsigned long, ip, FIELD_STRING_IP, 0);
@@ -1132,7 +1151,7 @@ static int kretprobe_event_define_fields(struct ftrace_event_call *event_call)
struct trace_probe *tp = (struct trace_probe *)event_call->data;
ret = trace_define_common_fields(event_call);
- if (!ret)
+ if (ret)
return ret;
DEFINE_FIELD(unsigned long, func, FIELD_STRING_FUNC, 0);
diff --git a/kernel/trace/trace_ksym.c b/kernel/trace/trace_ksym.c
index ddfa0fd..acb87d4 100644
--- a/kernel/trace/trace_ksym.c
+++ b/kernel/trace/trace_ksym.c
@@ -79,11 +79,12 @@ void ksym_collect_stats(unsigned long hbp_hit_addr)
}
#endif /* CONFIG_PROFILE_KSYM_TRACER */
-void ksym_hbp_handler(struct perf_event *hbp, void *data)
+void ksym_hbp_handler(struct perf_event *hbp, int nmi,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
{
struct ring_buffer_event *event;
struct ksym_trace_entry *entry;
- struct pt_regs *regs = data;
struct ring_buffer *buffer;
int pc;
diff --git a/samples/hw_breakpoint/data_breakpoint.c b/samples/hw_breakpoint/data_breakpoint.c
index 2952550..c69cbe9 100644
--- a/samples/hw_breakpoint/data_breakpoint.c
+++ b/samples/hw_breakpoint/data_breakpoint.c
@@ -41,7 +41,9 @@ module_param_string(ksym, ksym_name, KSYM_NAME_LEN, S_IRUGO);
MODULE_PARM_DESC(ksym, "Kernel symbol to monitor; this module will report any"
" write operations on the kernel symbol");
-static void sample_hbp_handler(struct perf_event *temp, void *data)
+static void sample_hbp_handler(struct perf_event *bp, int nmi,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
{
printk(KERN_INFO "%s value is changed\n", ksym_name);
dump_stack();
@@ -51,8 +53,9 @@ static void sample_hbp_handler(struct perf_event *temp, void *data)
static int __init hw_break_module_init(void)
{
int ret;
- DEFINE_BREAKPOINT_ATTR(attr);
+ struct perf_event_attr attr;
+ hw_breakpoint_init(&attr);
attr.bp_addr = kallsyms_lookup_name(ksym_name);
attr.bp_len = HW_BREAKPOINT_LEN_4;
attr.bp_type = HW_BREAKPOINT_W | HW_BREAKPOINT_R;
diff --git a/tools/perf/Documentation/perf-kmem.txt b/tools/perf/Documentation/perf-kmem.txt
index 44b0ce3..eac4d85 100644
--- a/tools/perf/Documentation/perf-kmem.txt
+++ b/tools/perf/Documentation/perf-kmem.txt
@@ -8,16 +8,16 @@ perf-kmem - Tool to trace/measure kernel memory(slab) properties
SYNOPSIS
--------
[verse]
-'perf kmem' {record} [<options>]
+'perf kmem' {record|stat} [<options>]
DESCRIPTION
-----------
-There's two variants of perf kmem:
+There are two variants of perf kmem:
'perf kmem record <command>' to record the kmem events
of an arbitrary workload.
- 'perf kmem' to report kernel memory statistics.
+ 'perf kmem stat' to report kernel memory statistics.
OPTIONS
-------
@@ -25,8 +25,11 @@ OPTIONS
--input=<file>::
Select the input file (default: perf.data)
---stat=<caller|alloc>::
- Select per callsite or per allocation statistics
+--caller::
+ Show per-callsite statistics
+
+--alloc::
+ Show per-allocation statistics
-s <key[,key2...]>::
--sort=<key[,key2...]>::
diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt
index 9270594..8fa6bf9 100644
--- a/tools/perf/Documentation/perf-probe.txt
+++ b/tools/perf/Documentation/perf-probe.txt
@@ -8,10 +8,13 @@ perf-probe - Define new dynamic tracepoints
SYNOPSIS
--------
[verse]
-'perf probe' [options] --add 'PROBE' [--add 'PROBE' ...]
+'perf probe' [options] --add='PROBE' [...]
or
-'perf probe' [options] 'PROBE' ['PROBE' ...]
-
+'perf probe' [options] PROBE
+or
+'perf probe' [options] --del='[GROUP:]EVENT' [...]
+or
+'perf probe' --list
DESCRIPTION
-----------
@@ -31,8 +34,16 @@ OPTIONS
Be more verbose (show parsed arguments, etc).
-a::
---add::
- Define a probe point (see PROBE SYNTAX for detail)
+--add=::
+ Define a probe event (see PROBE SYNTAX for detail).
+
+-d::
+--del=::
+ Delete a probe event.
+
+-l::
+--list::
+ List up current probe events.
PROBE SYNTAX
------------
diff --git a/tools/perf/builtin-buildid-list.c b/tools/perf/builtin-buildid-list.c
index 7dee9d1..dcb6143 100644
--- a/tools/perf/builtin-buildid-list.c
+++ b/tools/perf/builtin-buildid-list.c
@@ -19,7 +19,7 @@ static char const *input_name = "perf.data";
static int force;
static const char *const buildid_list_usage[] = {
- "perf report [<options>]",
+ "perf buildid-list [<options>]",
NULL
};
diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index 047fef7..5f20951 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -57,11 +57,6 @@ static struct rb_root root_caller_sorted;
static unsigned long total_requested, total_allocated;
static unsigned long nr_allocs, nr_cross_allocs;
-struct raw_event_sample {
- u32 size;
- char data[0];
-};
-
#define PATH_SYS_NODE "/sys/devices/system/node"
static void init_cpunode_map(void)
@@ -201,7 +196,7 @@ static void insert_caller_stat(unsigned long call_site,
}
}
-static void process_alloc_event(struct raw_event_sample *raw,
+static void process_alloc_event(void *data,
struct event *event,
int cpu,
u64 timestamp __used,
@@ -214,10 +209,10 @@ static void process_alloc_event(struct raw_event_sample *raw,
int bytes_alloc;
int node1, node2;
- ptr = raw_field_value(event, "ptr", raw->data);
- call_site = raw_field_value(event, "call_site", raw->data);
- bytes_req = raw_field_value(event, "bytes_req", raw->data);
- bytes_alloc = raw_field_value(event, "bytes_alloc", raw->data);
+ ptr = raw_field_value(event, "ptr", data);
+ call_site = raw_field_value(event, "call_site", data);
+ bytes_req = raw_field_value(event, "bytes_req", data);
+ bytes_alloc = raw_field_value(event, "bytes_alloc", data);
insert_alloc_stat(call_site, ptr, bytes_req, bytes_alloc, cpu);
insert_caller_stat(call_site, bytes_req, bytes_alloc);
@@ -227,7 +222,7 @@ static void process_alloc_event(struct raw_event_sample *raw,
if (node) {
node1 = cpunode_map[cpu];
- node2 = raw_field_value(event, "node", raw->data);
+ node2 = raw_field_value(event, "node", data);
if (node1 != node2)
nr_cross_allocs++;
}
@@ -262,7 +257,7 @@ static struct alloc_stat *search_alloc_stat(unsigned long ptr,
return NULL;
}
-static void process_free_event(struct raw_event_sample *raw,
+static void process_free_event(void *data,
struct event *event,
int cpu,
u64 timestamp __used,
@@ -271,7 +266,7 @@ static void process_free_event(struct raw_event_sample *raw,
unsigned long ptr;
struct alloc_stat *s_alloc, *s_caller;
- ptr = raw_field_value(event, "ptr", raw->data);
+ ptr = raw_field_value(event, "ptr", data);
s_alloc = search_alloc_stat(ptr, 0, &root_alloc_stat, ptr_cmp);
if (!s_alloc)
@@ -289,66 +284,53 @@ static void process_free_event(struct raw_event_sample *raw,
}
static void
-process_raw_event(event_t *raw_event __used, void *more_data,
+process_raw_event(event_t *raw_event __used, void *data,
int cpu, u64 timestamp, struct thread *thread)
{
- struct raw_event_sample *raw = more_data;
struct event *event;
int type;
- type = trace_parse_common_type(raw->data);
+ type = trace_parse_common_type(data);
event = trace_find_event(type);
if (!strcmp(event->name, "kmalloc") ||
!strcmp(event->name, "kmem_cache_alloc")) {
- process_alloc_event(raw, event, cpu, timestamp, thread, 0);
+ process_alloc_event(data, event, cpu, timestamp, thread, 0);
return;
}
if (!strcmp(event->name, "kmalloc_node") ||
!strcmp(event->name, "kmem_cache_alloc_node")) {
- process_alloc_event(raw, event, cpu, timestamp, thread, 1);
+ process_alloc_event(data, event, cpu, timestamp, thread, 1);
return;
}
if (!strcmp(event->name, "kfree") ||
!strcmp(event->name, "kmem_cache_free")) {
- process_free_event(raw, event, cpu, timestamp, thread);
+ process_free_event(data, event, cpu, timestamp, thread);
return;
}
}
static int process_sample_event(event_t *event)
{
- u64 ip = event->ip.ip;
- u64 timestamp = -1;
- u32 cpu = -1;
- u64 period = 1;
- void *more_data = event->ip.__more_data;
- struct thread *thread = threads__findnew(event->ip.pid);
+ struct sample_data data;
+ struct thread *thread;
- if (sample_type & PERF_SAMPLE_TIME) {
- timestamp = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
-
- if (sample_type & PERF_SAMPLE_CPU) {
- cpu = *(u32 *)more_data;
- more_data += sizeof(u32);
- more_data += sizeof(u32); /* reserved */
- }
+ memset(&data, 0, sizeof(data));
+ data.time = -1;
+ data.cpu = -1;
+ data.period = 1;
- if (sample_type & PERF_SAMPLE_PERIOD) {
- period = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
+ event__parse_sample(event, sample_type, &data);
dump_printf("(IP, %d): %d/%d: %p period: %Ld\n",
event->header.misc,
- event->ip.pid, event->ip.tid,
- (void *)(long)ip,
- (long long)period);
+ data.pid, data.tid,
+ (void *)(long)data.ip,
+ (long long)data.period);
+ thread = threads__findnew(event->ip.pid);
if (thread == NULL) {
pr_debug("problem processing %d event, skipping it.\n",
event->header.type);
@@ -357,7 +339,8 @@ static int process_sample_event(event_t *event)
dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
- process_raw_event(event, more_data, cpu, timestamp, thread);
+ process_raw_event(event, data.raw_data, data.cpu,
+ data.time, thread);
return 0;
}
@@ -543,7 +526,7 @@ static int __cmd_kmem(void)
}
static const char * const kmem_usage[] = {
- "perf kmem [<options>] {record}",
+ "perf kmem [<options>] {record|stat}",
NULL
};
@@ -703,18 +686,17 @@ static int parse_sort_opt(const struct option *opt __used,
return 0;
}
-static int parse_stat_opt(const struct option *opt __used,
- const char *arg, int unset __used)
+static int parse_caller_opt(const struct option *opt __used,
+ const char *arg __used, int unset __used)
{
- if (!arg)
- return -1;
+ caller_flag = (alloc_flag + 1);
+ return 0;
+}
- if (strcmp(arg, "alloc") == 0)
- alloc_flag = (caller_flag + 1);
- else if (strcmp(arg, "caller") == 0)
- caller_flag = (alloc_flag + 1);
- else
- return -1;
+static int parse_alloc_opt(const struct option *opt __used,
+ const char *arg __used, int unset __used)
+{
+ alloc_flag = (caller_flag + 1);
return 0;
}
@@ -739,14 +721,17 @@ static int parse_line_opt(const struct option *opt __used,
static const struct option kmem_options[] = {
OPT_STRING('i', "input", &input_name, "file",
"input file name"),
- OPT_CALLBACK(0, "stat", NULL, "<alloc>|<caller>",
- "stat selector, Pass 'alloc' or 'caller'.",
- parse_stat_opt),
+ OPT_CALLBACK_NOOPT(0, "caller", NULL, NULL,
+ "show per-callsite statistics",
+ parse_caller_opt),
+ OPT_CALLBACK_NOOPT(0, "alloc", NULL, NULL,
+ "show per-allocation statistics",
+ parse_alloc_opt),
OPT_CALLBACK('s', "sort", NULL, "key[,key2...]",
"sort by keys: ptr, call_site, bytes, hit, pingpong, frag",
parse_sort_opt),
OPT_CALLBACK('l', "line", NULL, "num",
- "show n lins",
+ "show n lines",
parse_line_opt),
OPT_BOOLEAN(0, "raw-ip", &raw_ip, "show raw ip instead of symbol"),
OPT_END()
@@ -790,18 +775,22 @@ int cmd_kmem(int argc, const char **argv, const char *prefix __used)
argc = parse_options(argc, argv, kmem_options, kmem_usage, 0);
- if (argc && !strncmp(argv[0], "rec", 3))
- return __cmd_record(argc, argv);
- else if (argc)
+ if (!argc)
usage_with_options(kmem_usage, kmem_options);
- if (list_empty(&caller_sort))
- setup_sorting(&caller_sort, default_sort_order);
- if (list_empty(&alloc_sort))
- setup_sorting(&alloc_sort, default_sort_order);
+ if (!strncmp(argv[0], "rec", 3)) {
+ return __cmd_record(argc, argv);
+ } else if (!strcmp(argv[0], "stat")) {
+ setup_cpunode_map();
+
+ if (list_empty(&caller_sort))
+ setup_sorting(&caller_sort, default_sort_order);
+ if (list_empty(&alloc_sort))
+ setup_sorting(&alloc_sort, default_sort_order);
- setup_cpunode_map();
+ return __cmd_kmem();
+ }
- return __cmd_kmem();
+ return 0;
}
diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c
index a58e11b..5a47c1e 100644
--- a/tools/perf/builtin-probe.c
+++ b/tools/perf/builtin-probe.c
@@ -35,6 +35,7 @@
#include "perf.h"
#include "builtin.h"
#include "util/util.h"
+#include "util/strlist.h"
#include "util/event.h"
#include "util/debug.h"
#include "util/parse-options.h"
@@ -43,11 +44,12 @@
#include "util/probe-event.h"
/* Default vmlinux search paths */
-#define NR_SEARCH_PATH 3
+#define NR_SEARCH_PATH 4
const char *default_search_path[NR_SEARCH_PATH] = {
"/lib/modules/%s/build/vmlinux", /* Custom build kernel */
"/usr/lib/debug/lib/modules/%s/vmlinux", /* Red Hat debuginfo */
"/boot/vmlinux-debug-%s", /* Ubuntu */
+"./vmlinux", /* CWD */
};
#define MAX_PATH_LEN 256
@@ -60,6 +62,7 @@ static struct {
int need_dwarf;
int nr_probe;
struct probe_point probes[MAX_PROBES];
+ struct strlist *dellist;
} session;
static bool listing;
@@ -79,6 +82,25 @@ static void parse_probe_event(const char *str)
pr_debug("%d arguments\n", pp->nr_args);
}
+static void parse_probe_event_argv(int argc, const char **argv)
+{
+ int i, len;
+ char *buf;
+
+ /* Bind up rest arguments */
+ len = 0;
+ for (i = 0; i < argc; i++)
+ len += strlen(argv[i]) + 1;
+ buf = zalloc(len + 1);
+ if (!buf)
+ die("Failed to allocate memory for binding arguments.");
+ len = 0;
+ for (i = 0; i < argc; i++)
+ len += sprintf(&buf[len], "%s ", argv[i]);
+ parse_probe_event(buf);
+ free(buf);
+}
+
static int opt_add_probe_event(const struct option *opt __used,
const char *str, int unset __used)
{
@@ -87,6 +109,17 @@ static int opt_add_probe_event(const struct option *opt __used,
return 0;
}
+static int opt_del_probe_event(const struct option *opt __used,
+ const char *str, int unset __used)
+{
+ if (str) {
+ if (!session.dellist)
+ session.dellist = strlist__new(true, NULL);
+ strlist__add(session.dellist, str);
+ }
+ return 0;
+}
+
#ifndef NO_LIBDWARF
static int open_default_vmlinux(void)
{
@@ -121,6 +154,7 @@ static int open_default_vmlinux(void)
static const char * const probe_usage[] = {
"perf probe [<options>] 'PROBEDEF' ['PROBEDEF' ...]",
"perf probe [<options>] --add 'PROBEDEF' [--add 'PROBEDEF' ...]",
+ "perf probe [<options>] --del '[GROUP:]EVENT' ...",
"perf probe --list",
NULL
};
@@ -132,7 +166,9 @@ static const struct option options[] = {
OPT_STRING('k', "vmlinux", &session.vmlinux, "file",
"vmlinux/module pathname"),
#endif
- OPT_BOOLEAN('l', "list", &listing, "list up current probes"),
+ OPT_BOOLEAN('l', "list", &listing, "list up current probe events"),
+ OPT_CALLBACK('d', "del", NULL, "[GROUP:]EVENT", "delete a probe event.",
+ opt_del_probe_event),
OPT_CALLBACK('a', "add", NULL,
#ifdef NO_LIBDWARF
"FUNC[+OFFS|%return] [ARG ...]",
@@ -160,7 +196,7 @@ static const struct option options[] = {
int cmd_probe(int argc, const char **argv, const char *prefix __used)
{
- int i, j, ret;
+ int i, ret;
#ifndef NO_LIBDWARF
int fd;
#endif
@@ -168,40 +204,52 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)
argc = parse_options(argc, argv, options, probe_usage,
PARSE_OPT_STOP_AT_NON_OPTION);
- for (i = 0; i < argc; i++)
- parse_probe_event(argv[i]);
+ if (argc > 0)
+ parse_probe_event_argv(argc, argv);
- if ((session.nr_probe == 0 && !listing) ||
- (session.nr_probe != 0 && listing))
+ if ((session.nr_probe == 0 && !session.dellist && !listing))
usage_with_options(probe_usage, options);
if (listing) {
+ if (session.nr_probe != 0 || session.dellist) {
+ pr_warning(" Error: Don't use --list with"
+ " --add/--del.\n");
+ usage_with_options(probe_usage, options);
+ }
show_perf_probe_events();
return 0;
}
+ if (session.dellist) {
+ del_trace_kprobe_events(session.dellist);
+ strlist__delete(session.dellist);
+ if (session.nr_probe == 0)
+ return 0;
+ }
+
if (session.need_dwarf)
#ifdef NO_LIBDWARF
die("Debuginfo-analysis is not supported");
#else /* !NO_LIBDWARF */
pr_debug("Some probes require debuginfo.\n");
- if (session.vmlinux)
+ if (session.vmlinux) {
+ pr_debug("Try to open %s.", session.vmlinux);
fd = open(session.vmlinux, O_RDONLY);
- else
+ } else
fd = open_default_vmlinux();
if (fd < 0) {
if (session.need_dwarf)
- die("Could not open vmlinux/module file.");
+ die("Could not open debuginfo file.");
- pr_warning("Could not open vmlinux/module file."
- " Try to use symbols.\n");
+ pr_debug("Could not open vmlinux/module file."
+ " Try to use symbols.\n");
goto end_dwarf;
}
/* Searching probe points */
- for (j = 0; j < session.nr_probe; j++) {
- pp = &session.probes[j];
+ for (i = 0; i < session.nr_probe; i++) {
+ pp = &session.probes[i];
if (pp->found)
continue;
@@ -223,8 +271,8 @@ end_dwarf:
#endif /* !NO_LIBDWARF */
/* Synthesize probes without dwarf */
- for (j = 0; j < session.nr_probe; j++) {
- pp = &session.probes[j];
+ for (i = 0; i < session.nr_probe; i++) {
+ pp = &session.probes[i];
if (pp->found) /* This probe is already found. */
continue;
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 383c4ab..2b9eb3a 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -605,44 +605,41 @@ static int validate_chain(struct ip_callchain *chain, event_t *event)
static int process_sample_event(event_t *event)
{
- u64 ip = event->ip.ip;
- u64 period = 1;
- void *more_data = event->ip.__more_data;
- struct ip_callchain *chain = NULL;
+ struct sample_data data;
int cpumode;
struct addr_location al;
- struct thread *thread = threads__findnew(event->ip.pid);
+ struct thread *thread;
- if (sample_type & PERF_SAMPLE_PERIOD) {
- period = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
+ memset(&data, 0, sizeof(data));
+ data.period = 1;
+
+ event__parse_sample(event, sample_type, &data);
dump_printf("(IP, %d): %d/%d: %p period: %Ld\n",
event->header.misc,
- event->ip.pid, event->ip.tid,
- (void *)(long)ip,
- (long long)period);
+ data.pid, data.tid,
+ (void *)(long)data.ip,
+ (long long)data.period);
if (sample_type & PERF_SAMPLE_CALLCHAIN) {
unsigned int i;
- chain = (void *)more_data;
-
- dump_printf("... chain: nr:%Lu\n", chain->nr);
+ dump_printf("... chain: nr:%Lu\n", data.callchain->nr);
- if (validate_chain(chain, event) < 0) {
+ if (validate_chain(data.callchain, event) < 0) {
pr_debug("call-chain problem with event, "
"skipping it.\n");
return 0;
}
if (dump_trace) {
- for (i = 0; i < chain->nr; i++)
- dump_printf("..... %2d: %016Lx\n", i, chain->ips[i]);
+ for (i = 0; i < data.callchain->nr; i++)
+ dump_printf("..... %2d: %016Lx\n",
+ i, data.callchain->ips[i]);
}
}
+ thread = threads__findnew(data.pid);
if (thread == NULL) {
pr_debug("problem processing %d event, skipping it.\n",
event->header.type);
@@ -657,7 +654,7 @@ static int process_sample_event(event_t *event)
cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
thread__find_addr_location(thread, cpumode,
- MAP__FUNCTION, ip, &al, NULL);
+ MAP__FUNCTION, data.ip, &al, NULL);
/*
* We have to do this here as we may have a dso with no symbol hit that
* has a name longer than the ones with symbols sampled.
@@ -675,12 +672,12 @@ static int process_sample_event(event_t *event)
if (sym_list && al.sym && !strlist__has_entry(sym_list, al.sym->name))
return 0;
- if (hist_entry__add(&al, chain, period)) {
+ if (hist_entry__add(&al, data.callchain, data.period)) {
pr_debug("problem incrementing symbol count, skipping event\n");
return -1;
}
- event__stats.total += period;
+ event__stats.total += data.period;
return 0;
}
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 26b782f..7cca7c1 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -13,7 +13,6 @@
#include "util/debug.h"
#include "util/data_map.h"
-#include <sys/types.h>
#include <sys/prctl.h>
#include <semaphore.h>
@@ -141,6 +140,7 @@ struct work_atoms {
struct thread *thread;
struct rb_node node;
u64 max_lat;
+ u64 max_lat_at;
u64 total_lat;
u64 nb_atoms;
u64 total_runtime;
@@ -414,34 +414,33 @@ static u64 get_cpu_usage_nsec_parent(void)
return sum;
}
-static u64 get_cpu_usage_nsec_self(void)
+static int self_open_counters(void)
{
- char filename [] = "/proc/1234567890/sched";
- unsigned long msecs, nsecs;
- char *line = NULL;
- u64 total = 0;
- size_t len = 0;
- ssize_t chars;
- FILE *file;
- int ret;
+ struct perf_event_attr attr;
+ int fd;
- sprintf(filename, "/proc/%d/sched", getpid());
- file = fopen(filename, "r");
- BUG_ON(!file);
+ memset(&attr, 0, sizeof(attr));
- while ((chars = getline(&line, &len, file)) != -1) {
- ret = sscanf(line, "se.sum_exec_runtime : %ld.%06ld\n",
- &msecs, &nsecs);
- if (ret == 2) {
- total = msecs*1e6 + nsecs;
- break;
- }
- }
- if (line)
- free(line);
- fclose(file);
+ attr.type = PERF_TYPE_SOFTWARE;
+ attr.config = PERF_COUNT_SW_TASK_CLOCK;
+
+ fd = sys_perf_event_open(&attr, 0, -1, -1, 0);
- return total;
+ if (fd < 0)
+ die("Error: sys_perf_event_open() syscall returned"
+ "with %d (%s)\n", fd, strerror(errno));
+ return fd;
+}
+
+static u64 get_cpu_usage_nsec_self(int fd)
+{
+ u64 runtime;
+ int ret;
+
+ ret = read(fd, &runtime, sizeof(runtime));
+ BUG_ON(ret != sizeof(runtime));
+
+ return runtime;
}
static void *thread_func(void *ctx)
@@ -450,9 +449,11 @@ static void *thread_func(void *ctx)
u64 cpu_usage_0, cpu_usage_1;
unsigned long i, ret;
char comm2[22];
+ int fd;
sprintf(comm2, ":%s", this_task->comm);
prctl(PR_SET_NAME, comm2);
+ fd = self_open_counters();
again:
ret = sem_post(&this_task->ready_for_work);
@@ -462,16 +463,15 @@ again:
ret = pthread_mutex_unlock(&start_work_mutex);
BUG_ON(ret);
- cpu_usage_0 = get_cpu_usage_nsec_self();
+ cpu_usage_0 = get_cpu_usage_nsec_self(fd);
for (i = 0; i < this_task->nr_events; i++) {
this_task->curr_event = i;
process_sched_event(this_task, this_task->atoms[i]);
}
- cpu_usage_1 = get_cpu_usage_nsec_self();
+ cpu_usage_1 = get_cpu_usage_nsec_self(fd);
this_task->cpu_usage = cpu_usage_1 - cpu_usage_0;
-
ret = sem_post(&this_task->work_done_sem);
BUG_ON(ret);
@@ -628,11 +628,6 @@ static void test_calibrations(void)
printf("the sleep test took %Ld nsecs\n", T1-T0);
}
-struct raw_event_sample {
- u32 size;
- char data[0];
-};
-
#define FILL_FIELD(ptr, field, event, data) \
ptr.field = (typeof(ptr.field)) raw_field_value(event, #field, data)
@@ -1019,8 +1014,10 @@ add_sched_in_event(struct work_atoms *atoms, u64 timestamp)
delta = atom->sched_in_time - atom->wake_up_time;
atoms->total_lat += delta;
- if (delta > atoms->max_lat)
+ if (delta > atoms->max_lat) {
atoms->max_lat = delta;
+ atoms->max_lat_at = timestamp;
+ }
atoms->nb_atoms++;
}
@@ -1216,10 +1213,11 @@ static void output_lat_thread(struct work_atoms *work_list)
avg = work_list->total_lat / work_list->nb_atoms;
- printf("|%11.3f ms |%9llu | avg:%9.3f ms | max:%9.3f ms |\n",
+ printf("|%11.3f ms |%9llu | avg:%9.3f ms | max:%9.3f ms | max at: %9.6f s\n",
(double)work_list->total_runtime / 1e6,
work_list->nb_atoms, (double)avg / 1e6,
- (double)work_list->max_lat / 1e6);
+ (double)work_list->max_lat / 1e6,
+ (double)work_list->max_lat_at / 1e9);
}
static int pid_cmp(struct work_atoms *l, struct work_atoms *r)
@@ -1356,7 +1354,7 @@ static void sort_lat(void)
static struct trace_sched_handler *trace_handler;
static void
-process_sched_wakeup_event(struct raw_event_sample *raw,
+process_sched_wakeup_event(void *data,
struct event *event,
int cpu __used,
u64 timestamp __used,
@@ -1364,13 +1362,13 @@ process_sched_wakeup_event(struct raw_event_sample *raw,
{
struct trace_wakeup_event wakeup_event;
- FILL_COMMON_FIELDS(wakeup_event, event, raw->data);
+ FILL_COMMON_FIELDS(wakeup_event, event, data);
- FILL_ARRAY(wakeup_event, comm, event, raw->data);
- FILL_FIELD(wakeup_event, pid, event, raw->data);
- FILL_FIELD(wakeup_event, prio, event, raw->data);
- FILL_FIELD(wakeup_event, success, event, raw->data);
- FILL_FIELD(wakeup_event, cpu, event, raw->data);
+ FILL_ARRAY(wakeup_event, comm, event, data);
+ FILL_FIELD(wakeup_event, pid, event, data);
+ FILL_FIELD(wakeup_event, prio, event, data);
+ FILL_FIELD(wakeup_event, success, event, data);
+ FILL_FIELD(wakeup_event, cpu, event, data);
if (trace_handler->wakeup_event)
trace_handler->wakeup_event(&wakeup_event, event, cpu, timestamp, thread);
@@ -1469,7 +1467,7 @@ map_switch_event(struct trace_switch_event *switch_event,
static void
-process_sched_switch_event(struct raw_event_sample *raw,
+process_sched_switch_event(void *data,
struct event *event,
int this_cpu,
u64 timestamp __used,
@@ -1477,15 +1475,15 @@ process_sched_switch_event(struct raw_event_sample *raw,
{
struct trace_switch_event switch_event;
- FILL_COMMON_FIELDS(switch_event, event, raw->data);
+ FILL_COMMON_FIELDS(switch_event, event, data);
- FILL_ARRAY(switch_event, prev_comm, event, raw->data);
- FILL_FIELD(switch_event, prev_pid, event, raw->data);
- FILL_FIELD(switch_event, prev_prio, event, raw->data);
- FILL_FIELD(switch_event, prev_state, event, raw->data);
- FILL_ARRAY(switch_event, next_comm, event, raw->data);
- FILL_FIELD(switch_event, next_pid, event, raw->data);
- FILL_FIELD(switch_event, next_prio, event, raw->data);
+ FILL_ARRAY(switch_event, prev_comm, event, data);
+ FILL_FIELD(switch_event, prev_pid, event, data);
+ FILL_FIELD(switch_event, prev_prio, event, data);
+ FILL_FIELD(switch_event, prev_state, event, data);
+ FILL_ARRAY(switch_event, next_comm, event, data);
+ FILL_FIELD(switch_event, next_pid, event, data);
+ FILL_FIELD(switch_event, next_prio, event, data);
if (curr_pid[this_cpu] != (u32)-1) {
/*
@@ -1502,7 +1500,7 @@ process_sched_switch_event(struct raw_event_sample *raw,
}
static void
-process_sched_runtime_event(struct raw_event_sample *raw,
+process_sched_runtime_event(void *data,
struct event *event,
int cpu __used,
u64 timestamp __used,
@@ -1510,17 +1508,17 @@ process_sched_runtime_event(struct raw_event_sample *raw,
{
struct trace_runtime_event runtime_event;
- FILL_ARRAY(runtime_event, comm, event, raw->data);
- FILL_FIELD(runtime_event, pid, event, raw->data);
- FILL_FIELD(runtime_event, runtime, event, raw->data);
- FILL_FIELD(runtime_event, vruntime, event, raw->data);
+ FILL_ARRAY(runtime_event, comm, event, data);
+ FILL_FIELD(runtime_event, pid, event, data);
+ FILL_FIELD(runtime_event, runtime, event, data);
+ FILL_FIELD(runtime_event, vruntime, event, data);
if (trace_handler->runtime_event)
trace_handler->runtime_event(&runtime_event, event, cpu, timestamp, thread);
}
static void
-process_sched_fork_event(struct raw_event_sample *raw,
+process_sched_fork_event(void *data,
struct event *event,
int cpu __used,
u64 timestamp __used,
@@ -1528,12 +1526,12 @@ process_sched_fork_event(struct raw_event_sample *raw,
{
struct trace_fork_event fork_event;
- FILL_COMMON_FIELDS(fork_event, event, raw->data);
+ FILL_COMMON_FIELDS(fork_event, event, data);
- FILL_ARRAY(fork_event, parent_comm, event, raw->data);
- FILL_FIELD(fork_event, parent_pid, event, raw->data);
- FILL_ARRAY(fork_event, child_comm, event, raw->data);
- FILL_FIELD(fork_event, child_pid, event, raw->data);
+ FILL_ARRAY(fork_event, parent_comm, event, data);
+ FILL_FIELD(fork_event, parent_pid, event, data);
+ FILL_ARRAY(fork_event, child_comm, event, data);
+ FILL_FIELD(fork_event, child_pid, event, data);
if (trace_handler->fork_event)
trace_handler->fork_event(&fork_event, event, cpu, timestamp, thread);
@@ -1550,7 +1548,7 @@ process_sched_exit_event(struct event *event,
}
static void
-process_sched_migrate_task_event(struct raw_event_sample *raw,
+process_sched_migrate_task_event(void *data,
struct event *event,
int cpu __used,
u64 timestamp __used,
@@ -1558,80 +1556,66 @@ process_sched_migrate_task_event(struct raw_event_sample *raw,
{
struct trace_migrate_task_event migrate_task_event;
- FILL_COMMON_FIELDS(migrate_task_event, event, raw->data);
+ FILL_COMMON_FIELDS(migrate_task_event, event, data);
- FILL_ARRAY(migrate_task_event, comm, event, raw->data);
- FILL_FIELD(migrate_task_event, pid, event, raw->data);
- FILL_FIELD(migrate_task_event, prio, event, raw->data);
- FILL_FIELD(migrate_task_event, cpu, event, raw->data);
+ FILL_ARRAY(migrate_task_event, comm, event, data);
+ FILL_FIELD(migrate_task_event, pid, event, data);
+ FILL_FIELD(migrate_task_event, prio, event, data);
+ FILL_FIELD(migrate_task_event, cpu, event, data);
if (trace_handler->migrate_task_event)
trace_handler->migrate_task_event(&migrate_task_event, event, cpu, timestamp, thread);
}
static void
-process_raw_event(event_t *raw_event __used, void *more_data,
+process_raw_event(event_t *raw_event __used, void *data,
int cpu, u64 timestamp, struct thread *thread)
{
- struct raw_event_sample *raw = more_data;
struct event *event;
int type;
- type = trace_parse_common_type(raw->data);
+
+ type = trace_parse_common_type(data);
event = trace_find_event(type);
if (!strcmp(event->name, "sched_switch"))
- process_sched_switch_event(raw, event, cpu, timestamp, thread);
+ process_sched_switch_event(data, event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_stat_runtime"))
- process_sched_runtime_event(raw, event, cpu, timestamp, thread);
+ process_sched_runtime_event(data, event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_wakeup"))
- process_sched_wakeup_event(raw, event, cpu, timestamp, thread);
+ process_sched_wakeup_event(data, event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_wakeup_new"))
- process_sched_wakeup_event(raw, event, cpu, timestamp, thread);
+ process_sched_wakeup_event(data, event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_process_fork"))
- process_sched_fork_event(raw, event, cpu, timestamp, thread);
+ process_sched_fork_event(data, event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_process_exit"))
process_sched_exit_event(event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_migrate_task"))
- process_sched_migrate_task_event(raw, event, cpu, timestamp, thread);
+ process_sched_migrate_task_event(data, event, cpu, timestamp, thread);
}
static int process_sample_event(event_t *event)
{
+ struct sample_data data;
struct thread *thread;
- u64 ip = event->ip.ip;
- u64 timestamp = -1;
- u32 cpu = -1;
- u64 period = 1;
- void *more_data = event->ip.__more_data;
if (!(sample_type & PERF_SAMPLE_RAW))
return 0;
- thread = threads__findnew(event->ip.pid);
+ memset(&data, 0, sizeof(data));
+ data.time = -1;
+ data.cpu = -1;
+ data.period = -1;
- if (sample_type & PERF_SAMPLE_TIME) {
- timestamp = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
-
- if (sample_type & PERF_SAMPLE_CPU) {
- cpu = *(u32 *)more_data;
- more_data += sizeof(u32);
- more_data += sizeof(u32); /* reserved */
- }
-
- if (sample_type & PERF_SAMPLE_PERIOD) {
- period = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
+ event__parse_sample(event, sample_type, &data);
dump_printf("(IP, %d): %d/%d: %p period: %Ld\n",
event->header.misc,
- event->ip.pid, event->ip.tid,
- (void *)(long)ip,
- (long long)period);
+ data.pid, data.tid,
+ (void *)(long)data.ip,
+ (long long)data.period);
+ thread = threads__findnew(data.pid);
if (thread == NULL) {
pr_debug("problem processing %d event, skipping it.\n",
event->header.type);
@@ -1640,10 +1624,10 @@ static int process_sample_event(event_t *event)
dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
- if (profile_cpu != -1 && profile_cpu != (int) cpu)
+ if (profile_cpu != -1 && profile_cpu != (int)data.cpu)
return 0;
- process_raw_event(event, more_data, cpu, timestamp, thread);
+ process_raw_event(event, data.raw_data, data.cpu, data.time, thread);
return 0;
}
@@ -1724,9 +1708,9 @@ static void __cmd_lat(void)
read_events();
sort_lat();
- printf("\n -----------------------------------------------------------------------------------------\n");
- printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms |\n");
- printf(" -----------------------------------------------------------------------------------------\n");
+ printf("\n ---------------------------------------------------------------------------------------------------------------\n");
+ printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms | Maximum delay at |\n");
+ printf(" ---------------------------------------------------------------------------------------------------------------\n");
next = rb_first(&sorted_atom_root);
@@ -1902,13 +1886,18 @@ static int __cmd_record(int argc, const char **argv)
int cmd_sched(int argc, const char **argv, const char *prefix __used)
{
- symbol__init(0);
-
argc = parse_options(argc, argv, sched_options, sched_usage,
PARSE_OPT_STOP_AT_NON_OPTION);
if (!argc)
usage_with_options(sched_usage, sched_options);
+ /*
+ * Aliased to 'perf trace' for now:
+ */
+ if (!strcmp(argv[0], "trace"))
+ return cmd_trace(argc, argv, prefix);
+
+ symbol__init(0);
if (!strncmp(argv[0], "rec", 3)) {
return __cmd_record(argc, argv);
} else if (!strncmp(argv[0], "lat", 3)) {
@@ -1932,11 +1921,6 @@ int cmd_sched(int argc, const char **argv, const char *prefix __used)
usage_with_options(replay_usage, replay_options);
}
__cmd_replay();
- } else if (!strcmp(argv[0], "trace")) {
- /*
- * Aliased to 'perf trace' for now:
- */
- return cmd_trace(argc, argv, prefix);
} else {
usage_with_options(sched_usage, sched_options);
}
diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c
index cb58b66..f472df9 100644
--- a/tools/perf/builtin-timechart.c
+++ b/tools/perf/builtin-timechart.c
@@ -302,12 +302,11 @@ process_exit_event(event_t *event)
}
struct trace_entry {
- u32 size;
unsigned short type;
unsigned char flags;
unsigned char preempt_count;
int pid;
- int tgid;
+ int lock_depth;
};
struct power_entry {
@@ -484,43 +483,22 @@ static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te)
static int
process_sample_event(event_t *event)
{
- int cursor = 0;
- u64 addr = 0;
- u64 stamp = 0;
- u32 cpu = 0;
- u32 pid = 0;
+ struct sample_data data;
struct trace_entry *te;
- if (sample_type & PERF_SAMPLE_IP)
- cursor++;
-
- if (sample_type & PERF_SAMPLE_TID) {
- pid = event->sample.array[cursor]>>32;
- cursor++;
- }
- if (sample_type & PERF_SAMPLE_TIME) {
- stamp = event->sample.array[cursor++];
+ memset(&data, 0, sizeof(data));
- if (!first_time || first_time > stamp)
- first_time = stamp;
- if (last_time < stamp)
- last_time = stamp;
+ event__parse_sample(event, sample_type, &data);
+ if (sample_type & PERF_SAMPLE_TIME) {
+ if (!first_time || first_time > data.time)
+ first_time = data.time;
+ if (last_time < data.time)
+ last_time = data.time;
}
- if (sample_type & PERF_SAMPLE_ADDR)
- addr = event->sample.array[cursor++];
- if (sample_type & PERF_SAMPLE_ID)
- cursor++;
- if (sample_type & PERF_SAMPLE_STREAM_ID)
- cursor++;
- if (sample_type & PERF_SAMPLE_CPU)
- cpu = event->sample.array[cursor++] & 0xFFFFFFFF;
- if (sample_type & PERF_SAMPLE_PERIOD)
- cursor++;
-
- te = (void *)&event->sample.array[cursor];
- if (sample_type & PERF_SAMPLE_RAW && te->size > 0) {
+ te = (void *)data.raw_data;
+ if (sample_type & PERF_SAMPLE_RAW && data.raw_size > 0) {
char *event_str;
struct power_entry *pe;
@@ -532,19 +510,19 @@ process_sample_event(event_t *event)
return 0;
if (strcmp(event_str, "power:power_start") == 0)
- c_state_start(cpu, stamp, pe->value);
+ c_state_start(data.cpu, data.time, pe->value);
if (strcmp(event_str, "power:power_end") == 0)
- c_state_end(cpu, stamp);
+ c_state_end(data.cpu, data.time);
if (strcmp(event_str, "power:power_frequency") == 0)
- p_state_change(cpu, stamp, pe->value);
+ p_state_change(data.cpu, data.time, pe->value);
if (strcmp(event_str, "sched:sched_wakeup") == 0)
- sched_wakeup(cpu, stamp, pid, te);
+ sched_wakeup(data.cpu, data.time, data.pid, te);
if (strcmp(event_str, "sched:sched_switch") == 0)
- sched_switch(cpu, stamp, te);
+ sched_switch(data.cpu, data.time, te);
}
return 0;
}
diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index abb914a..c2fcc34 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -66,58 +66,40 @@ static u64 sample_type;
static int process_sample_event(event_t *event)
{
- u64 ip = event->ip.ip;
- u64 timestamp = -1;
- u32 cpu = -1;
- u64 period = 1;
- void *more_data = event->ip.__more_data;
- struct thread *thread = threads__findnew(event->ip.pid);
-
- if (sample_type & PERF_SAMPLE_TIME) {
- timestamp = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
+ struct sample_data data;
+ struct thread *thread;
- if (sample_type & PERF_SAMPLE_CPU) {
- cpu = *(u32 *)more_data;
- more_data += sizeof(u32);
- more_data += sizeof(u32); /* reserved */
- }
+ memset(&data, 0, sizeof(data));
+ data.time = -1;
+ data.cpu = -1;
+ data.period = 1;
- if (sample_type & PERF_SAMPLE_PERIOD) {
- period = *(u64 *)more_data;
- more_data += sizeof(u64);
- }
+ event__parse_sample(event, sample_type, &data);
dump_printf("(IP, %d): %d/%d: %p period: %Ld\n",
event->header.misc,
- event->ip.pid, event->ip.tid,
- (void *)(long)ip,
- (long long)period);
+ data.pid, data.tid,
+ (void *)(long)data.ip,
+ (long long)data.period);
+ thread = threads__findnew(event->ip.pid);
if (thread == NULL) {
pr_debug("problem processing %d event, skipping it.\n",
event->header.type);
return -1;
}
- dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
-
if (sample_type & PERF_SAMPLE_RAW) {
- struct {
- u32 size;
- char data[0];
- } *raw = more_data;
-
/*
* FIXME: better resolve from pid from the struct trace_entry
* field, although it should be the same than this perf
* event pid
*/
- scripting_ops->process_event(cpu, raw->data, raw->size,
- timestamp, thread->comm);
+ scripting_ops->process_event(data.cpu, data.raw_data,
+ data.raw_size,
+ data.time, thread->comm);
}
- event__stats.total += period;
+ event__stats.total += data.period;
return 0;
}
diff --git a/tools/perf/util/data_map.c b/tools/perf/util/data_map.c
index ca0bedf..59b65d0 100644
--- a/tools/perf/util/data_map.c
+++ b/tools/perf/util/data_map.c
@@ -100,11 +100,11 @@ process_event(event_t *event, unsigned long offset, unsigned long head)
}
}
-int perf_header__read_build_ids(int input, off_t offset, off_t size)
+int perf_header__read_build_ids(int input, u64 offset, u64 size)
{
struct build_id_event bev;
char filename[PATH_MAX];
- off_t limit = offset + size;
+ u64 limit = offset + size;
int err = -1;
while (offset < limit) {
diff --git a/tools/perf/util/data_map.h b/tools/perf/util/data_map.h
index 3180ff7..258a87b 100644
--- a/tools/perf/util/data_map.h
+++ b/tools/perf/util/data_map.h
@@ -27,6 +27,6 @@ int mmap_dispatch_perf_file(struct perf_header **pheader,
int full_paths,
int *cwdlen,
char **cwd);
-int perf_header__read_build_ids(int input, off_t offset, off_t file_size);
+int perf_header__read_build_ids(int input, u64 offset, u64 file_size);
#endif
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 414b89d..4dcecaf 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -310,3 +310,70 @@ int event__preprocess_sample(const event_t *self, struct addr_location *al,
al->level == 'H' ? "[hypervisor]" : "<not found>");
return 0;
}
+
+int event__parse_sample(event_t *event, u64 type, struct sample_data *data)
+{
+ u64 *array = event->sample.array;
+
+ if (type & PERF_SAMPLE_IP) {
+ data->ip = event->ip.ip;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_TID) {
+ u32 *p = (u32 *)array;
+ data->pid = p[0];
+ data->tid = p[1];
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_TIME) {
+ data->time = *array;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_ADDR) {
+ data->addr = *array;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_ID) {
+ data->id = *array;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_STREAM_ID) {
+ data->stream_id = *array;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_CPU) {
+ u32 *p = (u32 *)array;
+ data->cpu = *p;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_PERIOD) {
+ data->period = *array;
+ array++;
+ }
+
+ if (type & PERF_SAMPLE_READ) {
+ pr_debug("PERF_SAMPLE_READ is unsuported for now\n");
+ return -1;
+ }
+
+ if (type & PERF_SAMPLE_CALLCHAIN) {
+ data->callchain = (struct ip_callchain *)array;
+ array += 1 + data->callchain->nr;
+ }
+
+ if (type & PERF_SAMPLE_RAW) {
+ u32 *p = (u32 *)array;
+ data->raw_size = *p;
+ p++;
+ data->raw_data = p;
+ }
+
+ return 0;
+}
diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
index a4cc810..c7a78ee 100644
--- a/tools/perf/util/event.h
+++ b/tools/perf/util/event.h
@@ -56,11 +56,25 @@ struct read_event {
u64 id;
};
-struct sample_event{
+struct sample_event {
struct perf_event_header header;
u64 array[];
};
+struct sample_data {
+ u64 ip;
+ u32 pid, tid;
+ u64 time;
+ u64 addr;
+ u64 id;
+ u64 stream_id;
+ u32 cpu;
+ u64 period;
+ struct ip_callchain *callchain;
+ u32 raw_size;
+ void *raw_data;
+};
+
#define BUILD_ID_SIZE 20
struct build_id_event {
@@ -155,5 +169,6 @@ int event__process_task(event_t *self);
struct addr_location;
int event__preprocess_sample(const event_t *self, struct addr_location *al,
symbol_filter_t filter);
+int event__parse_sample(event_t *event, u64 type, struct sample_data *data);
#endif /* __PERF_RECORD_H */
--git a/tools/perf/util/header.c b/tools/perf/util/header.c
index 4805e6d..59a9c0b 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -187,7 +187,9 @@ static int do_write(int fd, const void *buf, size_t size)
static int __dsos__write_buildid_table(struct list_head *head, int fd)
{
+#define NAME_ALIGN 64
struct dso *pos;
+ static const char zero_buf[NAME_ALIGN];
list_for_each_entry(pos, head, node) {
int err;
@@ -197,14 +199,17 @@ static int __dsos__write_buildid_table(struct list_head *head, int fd)
if (!pos->has_build_id)
continue;
len = pos->long_name_len + 1;
- len = ALIGN(len, 64);
+ len = ALIGN(len, NAME_ALIGN);
memset(&b, 0, sizeof(b));
memcpy(&b.build_id, pos->build_id, sizeof(pos->build_id));
b.header.size = sizeof(b) + len;
err = do_write(fd, &b, sizeof(b));
if (err < 0)
return err;
- err = do_write(fd, pos->long_name, len);
+ err = do_write(fd, pos->long_name, pos->long_name_len + 1);
+ if (err < 0)
+ return err;
+ err = do_write(fd, zero_buf, len - pos->long_name_len - 1);
if (err < 0)
return err;
}
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index 9e5dbd6..e5bc0fb 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -197,7 +197,7 @@ struct tracepoint_path *tracepoint_id_to_path(u64 config)
if (id == config) {
closedir(evt_dir);
closedir(sys_dir);
- path = zalloc(sizeof(path));
+ path = zalloc(sizeof(*path));
path->system = malloc(MAX_EVENT_LENGTH);
if (!path->system) {
free(path);
@@ -467,7 +467,6 @@ parse_subsystem_tracepoint_event(char *sys_name, char *flags)
while ((evt_ent = readdir(evt_dir))) {
char event_opt[MAX_EVOPT_LEN + 1];
int len;
- unsigned int rem = MAX_EVOPT_LEN;
if (!strcmp(evt_ent->d_name, ".")
|| !strcmp(evt_ent->d_name, "..")
@@ -475,20 +474,12 @@ parse_subsystem_tracepoint_event(char *sys_name, char *flags)
|| !strcmp(evt_ent->d_name, "filter"))
continue;
- len = snprintf(event_opt, MAX_EVOPT_LEN, "%s:%s", sys_name,
- evt_ent->d_name);
+ len = snprintf(event_opt, MAX_EVOPT_LEN, "%s:%s%s%s", sys_name,
+ evt_ent->d_name, flags ? ":" : "",
+ flags ?: "");
if (len < 0)
return EVT_FAILED;
- rem -= len;
- if (flags) {
- if (rem < strlen(flags) + 1)
- return EVT_FAILED;
-
- strcat(event_opt, ":");
- strcat(event_opt, flags);
- }
-
if (parse_events(NULL, event_opt, 0))
return EVT_FAILED;
}
diff --git a/tools/perf/util/parse-options.c b/tools/perf/util/parse-options.c
index 6d8af48..efebd5b 100644
--- a/tools/perf/util/parse-options.c
+++ b/tools/perf/util/parse-options.c
@@ -430,6 +430,9 @@ int usage_with_options_internal(const char * const *usagestr,
pos = fprintf(stderr, " ");
if (opts->short_name)
pos += fprintf(stderr, "-%c", opts->short_name);
+ else
+ pos += fprintf(stderr, " ");
+
if (opts->long_name && opts->short_name)
pos += fprintf(stderr, ", ");
if (opts->long_name)
diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index cd7fbda..d14a458 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -48,6 +48,9 @@
/* If there is no space to write, returns -E2BIG. */
static int e_snprintf(char *str, size_t size, const char *format, ...)
+ __attribute__((format(printf, 3, 4)));
+
+static int e_snprintf(char *str, size_t size, const char *format, ...)
{
int ret;
va_list ap;
@@ -258,7 +261,7 @@ int synthesize_perf_probe_event(struct probe_point *pp)
ret = e_snprintf(buf, MAX_CMDLEN, "%s%s%s%s", pp->function,
offs, pp->retprobe ? "%return" : "", line);
else
- ret = e_snprintf(buf, MAX_CMDLEN, "%s%s%s%s", pp->file, line);
+ ret = e_snprintf(buf, MAX_CMDLEN, "%s%s", pp->file, line);
if (ret <= 0)
goto error;
len = ret;
@@ -373,14 +376,32 @@ static void clear_probe_point(struct probe_point *pp)
free(pp->args);
for (i = 0; i < pp->found; i++)
free(pp->probes[i]);
- memset(pp, 0, sizeof(pp));
+ memset(pp, 0, sizeof(*pp));
+}
+
+/* Show an event */
+static void show_perf_probe_event(const char *group, const char *event,
+ const char *place, struct probe_point *pp)
+{
+ int i;
+ char buf[128];
+
+ e_snprintf(buf, 128, "%s:%s", group, event);
+ printf(" %-40s (on %s", buf, place);
+
+ if (pp->nr_args > 0) {
+ printf(" with");
+ for (i = 0; i < pp->nr_args; i++)
+ printf(" %s", pp->args[i]);
+ }
+ printf(")\n");
}
/* List up current perf-probe events */
void show_perf_probe_events(void)
{
unsigned int i;
- int fd;
+ int fd, nr;
char *group, *event;
struct probe_point pp;
struct strlist *rawlist;
@@ -393,8 +414,13 @@ void show_perf_probe_events(void)
for (i = 0; i < strlist__nr_entries(rawlist); i++) {
ent = strlist__entry(rawlist, i);
parse_trace_kprobe_event(ent->s, &group, &event, &pp);
+ /* Synthesize only event probe point */
+ nr = pp.nr_args;
+ pp.nr_args = 0;
synthesize_perf_probe_event(&pp);
- printf("[%s:%s]\t%s\n", group, event, pp.probes[0]);
+ pp.nr_args = nr;
+ /* Show an event */
+ show_perf_probe_event(group, event, pp.probes[0], &pp);
free(group);
free(event);
clear_probe_point(&pp);
@@ -404,21 +430,28 @@ void show_perf_probe_events(void)
}
/* Get current perf-probe event names */
-static struct strlist *get_perf_event_names(int fd)
+static struct strlist *get_perf_event_names(int fd, bool include_group)
{
unsigned int i;
char *group, *event;
+ char buf[128];
struct strlist *sl, *rawlist;
struct str_node *ent;
rawlist = get_trace_kprobe_event_rawlist(fd);
- sl = strlist__new(false, NULL);
+ sl = strlist__new(true, NULL);
for (i = 0; i < strlist__nr_entries(rawlist); i++) {
ent = strlist__entry(rawlist, i);
parse_trace_kprobe_event(ent->s, &group, &event, NULL);
- strlist__add(sl, event);
+ if (include_group) {
+ if (e_snprintf(buf, 128, "%s:%s", group, event) < 0)
+ die("Failed to copy group:event name.");
+ strlist__add(sl, buf);
+ } else
+ strlist__add(sl, event);
free(group);
+ free(event);
}
strlist__delete(rawlist);
@@ -426,24 +459,30 @@ static struct strlist *get_perf_event_names(int fd)
return sl;
}
-static int write_trace_kprobe_event(int fd, const char *buf)
+static void write_trace_kprobe_event(int fd, const char *buf)
{
int ret;
+ pr_debug("Writing event: %s\n", buf);
ret = write(fd, buf, strlen(buf));
if (ret <= 0)
- die("Failed to create event.");
- else
- printf("Added new event: %s\n", buf);
-
- return ret;
+ die("Failed to write event: %s", strerror(errno));
}
static void get_new_event_name(char *buf, size_t len, const char *base,
struct strlist *namelist)
{
int i, ret;
- for (i = 0; i < MAX_EVENT_INDEX; i++) {
+
+ /* Try no suffix */
+ ret = e_snprintf(buf, len, "%s", base);
+ if (ret < 0)
+ die("snprintf() failed: %s", strerror(-ret));
+ if (!strlist__has_entry(namelist, buf))
+ return;
+
+ /* Try to add suffix */
+ for (i = 1; i < MAX_EVENT_INDEX; i++) {
ret = e_snprintf(buf, len, "%s_%d", base, i);
if (ret < 0)
die("snprintf() failed: %s", strerror(-ret));
@@ -464,7 +503,7 @@ void add_trace_kprobe_events(struct probe_point *probes, int nr_probes)
fd = open_kprobe_events(O_RDWR, O_APPEND);
/* Get current event names */
- namelist = get_perf_event_names(fd);
+ namelist = get_perf_event_names(fd, false);
for (j = 0; j < nr_probes; j++) {
pp = probes + j;
@@ -476,9 +515,73 @@ void add_trace_kprobe_events(struct probe_point *probes, int nr_probes)
PERFPROBE_GROUP, event,
pp->probes[i]);
write_trace_kprobe_event(fd, buf);
+ printf("Added new event:\n");
+ /* Get the first parameter (probe-point) */
+ sscanf(pp->probes[i], "%s", buf);
+ show_perf_probe_event(PERFPROBE_GROUP, event,
+ buf, pp);
/* Add added event name to namelist */
strlist__add(namelist, event);
}
}
+ /* Show how to use the event. */
+ printf("\nYou can now use it on all perf tools, such as:\n\n");
+ printf("\tperf record -e %s:%s -a sleep 1\n\n", PERFPROBE_GROUP, event);
+
+ strlist__delete(namelist);
+ close(fd);
+}
+
+static void del_trace_kprobe_event(int fd, const char *group,
+ const char *event, struct strlist *namelist)
+{
+ char buf[128];
+
+ if (e_snprintf(buf, 128, "%s:%s", group, event) < 0)
+ die("Failed to copy event.");
+ if (!strlist__has_entry(namelist, buf)) {
+ pr_warning("Warning: event \"%s\" is not found.\n", buf);
+ return;
+ }
+ /* Convert from perf-probe event to trace-kprobe event */
+ if (e_snprintf(buf, 128, "-:%s/%s", group, event) < 0)
+ die("Failed to copy event.");
+
+ write_trace_kprobe_event(fd, buf);
+ printf("Remove event: %s:%s\n", group, event);
+}
+
+void del_trace_kprobe_events(struct strlist *dellist)
+{
+ int fd;
+ unsigned int i;
+ const char *group, *event;
+ char *p, *str;
+ struct str_node *ent;
+ struct strlist *namelist;
+
+ fd = open_kprobe_events(O_RDWR, O_APPEND);
+ /* Get current event names */
+ namelist = get_perf_event_names(fd, true);
+
+ for (i = 0; i < strlist__nr_entries(dellist); i++) {
+ ent = strlist__entry(dellist, i);
+ str = strdup(ent->s);
+ if (!str)
+ die("Failed to copy event.");
+ p = strchr(str, ':');
+ if (p) {
+ group = str;
+ *p = '\0';
+ event = p + 1;
+ } else {
+ group = PERFPROBE_GROUP;
+ event = str;
+ }
+ del_trace_kprobe_event(fd, group, event, namelist);
+ free(str);
+ }
+ strlist__delete(namelist);
close(fd);
}
+
diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h
index 0c6fe56..f752159 100644
--- a/tools/perf/util/probe-event.h
+++ b/tools/perf/util/probe-event.h
@@ -10,6 +10,7 @@ extern void parse_trace_kprobe_event(const char *str, char **group,
char **event, struct probe_point *pp);
extern int synthesize_trace_kprobe_event(struct probe_point *pp);
extern void add_trace_kprobe_events(struct probe_point *probes, int nr_probes);
+extern void del_trace_kprobe_events(struct strlist *dellist);
extern void show_perf_probe_events(void);
/* Maximum index number of event-name postfix */
diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c
index 293cdfc..4585f1d 100644
--- a/tools/perf/util/probe-finder.c
+++ b/tools/perf/util/probe-finder.c
@@ -106,7 +106,7 @@ static int strtailcmp(const char *s1, const char *s2)
{
int i1 = strlen(s1);
int i2 = strlen(s2);
- while (--i1 > 0 && --i2 > 0) {
+ while (--i1 >= 0 && --i2 >= 0) {
if (s1[i1] != s2[i2])
return s1[i1] - s2[i2];
}
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index fffcb93..e7508ad 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -938,8 +938,9 @@ static bool __dsos__read_build_ids(struct list_head *head)
bool dsos__read_build_ids(void)
{
- return __dsos__read_build_ids(&dsos__kernel) ||
- __dsos__read_build_ids(&dsos__user);
+ bool kbuildids = __dsos__read_build_ids(&dsos__kernel),
+ ubuildids = __dsos__read_build_ids(&dsos__user);
+ return kbuildids || ubuildids;
}
/*
diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c
index 0302405..c5c32be 100644
--- a/tools/perf/util/trace-event-parse.c
+++ b/tools/perf/util/trace-event-parse.c
@@ -177,7 +177,7 @@ void parse_proc_kallsyms(char *file, unsigned int size __unused)
func_count++;
}
- func_list = malloc_or_die(sizeof(*func_list) * func_count + 1);
+ func_list = malloc_or_die(sizeof(*func_list) * (func_count + 1));
i = 0;
while (list) {
@@ -1477,7 +1477,7 @@ process_fields(struct event *event, struct print_flag_sym **list, char **tok)
goto out_free;
field = malloc_or_die(sizeof(*field));
- memset(field, 0, sizeof(field));
+ memset(field, 0, sizeof(*field));
value = arg_eval(arg);
field->value = strdup(value);
diff --git a/tools/perf/util/trace-event-perl.c b/tools/perf/util/trace-event-perl.c
index 51e833f..a5ffe60 100644
--- a/tools/perf/util/trace-event-perl.c
+++ b/tools/perf/util/trace-event-perl.c
@@ -32,9 +32,6 @@
void xs_init(pTHX);
-void boot_Perf__Trace__Context(pTHX_ CV *cv);
-void boot_DynaLoader(pTHX_ CV *cv);
-
void xs_init(pTHX)
{
const char *file = __FILE__;
@@ -573,26 +570,72 @@ struct scripting_ops perl_scripting_ops = {
.generate_script = perl_generate_script,
};
-#ifdef NO_LIBPERL
-void setup_perl_scripting(void)
+static void print_unsupported_msg(void)
{
fprintf(stderr, "Perl scripting not supported."
- " Install libperl and rebuild perf to enable it. e.g. "
- "apt-get install libperl-dev (ubuntu), yum install "
- "perl-ExtUtils-Embed (Fedora), etc.\n");
+ " Install libperl and rebuild perf to enable it.\n"
+ "For example:\n # apt-get install libperl-dev (ubuntu)"
+ "\n # yum install perl-ExtUtils-Embed (Fedora)"
+ "\n etc.\n");
}
-#else
-void setup_perl_scripting(void)
+
+static int perl_start_script_unsupported(const char *script __unused)
+{
+ print_unsupported_msg();
+
+ return -1;
+}
+
+static int perl_stop_script_unsupported(void)
+{
+ return 0;
+}
+
+static void perl_process_event_unsupported(int cpu __unused,
+ void *data __unused,
+ int size __unused,
+ unsigned long long nsecs __unused,
+ char *comm __unused)
+{
+}
+
+static int perl_generate_script_unsupported(const char *outfile __unused)
+{
+ print_unsupported_msg();
+
+ return -1;
+}
+
+struct scripting_ops perl_scripting_unsupported_ops = {
+ .name = "Perl",
+ .start_script = perl_start_script_unsupported,
+ .stop_script = perl_stop_script_unsupported,
+ .process_event = perl_process_event_unsupported,
+ .generate_script = perl_generate_script_unsupported,
+};
+
+static void register_perl_scripting(struct scripting_ops *scripting_ops)
{
int err;
- err = script_spec_register("Perl", &perl_scripting_ops);
+ err = script_spec_register("Perl", scripting_ops);
if (err)
die("error registering Perl script extension");
- err = script_spec_register("pl", &perl_scripting_ops);
+ err = script_spec_register("pl", scripting_ops);
if (err)
die("error registering pl script extension");
scripting_context = malloc(sizeof(struct scripting_context));
}
+
+#ifdef NO_LIBPERL
+void setup_perl_scripting(void)
+{
+ register_perl_scripting(&perl_scripting_unsupported_ops);
+}
+#else
+void setup_perl_scripting(void)
+{
+ register_perl_scripting(&perl_scripting_ops);
+}
#endif
diff --git a/tools/perf/util/trace-event-perl.h b/tools/perf/util/trace-event-perl.h
index 8fe0d86..e88fb26 100644
--- a/tools/perf/util/trace-event-perl.h
+++ b/tools/perf/util/trace-event-perl.h
@@ -34,9 +34,13 @@ typedef int INTERP;
#define dXSUB_SYS
#define pTHX_
static inline void newXS(const char *a, void *b, const char *c) {}
+static void boot_Perf__Trace__Context(pTHX_ CV *cv) {}
+static void boot_DynaLoader(pTHX_ CV *cv) {}
#else
#include <EXTERN.h>
#include <perl.h>
+void boot_Perf__Trace__Context(pTHX_ CV *cv);
+void boot_DynaLoader(pTHX_ CV *cv);
typedef PerlInterpreter * INTERP;
#endif
diff --git a/tools/perf/util/trace-event-read.c b/tools/perf/util/trace-event-read.c
index 342dfdd..1744422 100644
--- a/tools/perf/util/trace-event-read.c
+++ b/tools/perf/util/trace-event-read.c
@@ -145,8 +145,9 @@ static void read_proc_kallsyms(void)
if (!size)
return;
- buf = malloc_or_die(size);
+ buf = malloc_or_die(size + 1);
read_or_die(buf, size);
+ buf[size] = '\0';
parse_proc_kallsyms(buf, size);
^ permalink raw reply related [flat|nested] 8+ messages in thread
end of thread, other threads:[~2010-01-29 7:07 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-01-29 2:07 [GIT PULL] perf events updates Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 1/5] perf: Fix inconsistency between IP and callchain sampling Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 2/5] perf: Factorize trace events raw sample buffer operations Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 3/5] tracing/kprobe: Cleanup unused return value of tracing functions Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 4/5] x86/debug: Clear reserved bits of DR6 in do_debug() Frederic Weisbecker
2010-01-29 2:07 ` [PATCH 5/5] x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler Frederic Weisbecker
2010-01-29 7:06 ` [GIT PULL] perf events updates Ingo Molnar
-- strict thread matches above, loose matches on Subject: below --
2009-12-10 19:25 Ingo Molnar
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).