[PATCH tracing/kprobes 1/3] tracing/kprobes: Use global event profile buffer in kprobe tracer

[PATCH tracing/kprobes 1/3] tracing/kprobes: Use global event profile buffer in kprobe tracer

[Masami Hiramatsu]

Use new percpu global event profile buffer instead of stack in kprobe
tracer.

Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Frank Ch. Eigler <fche@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Jason Baron <jbaron@redhat.com>
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
---

 kernel/trace/trace_kprobe.c |  115 +++++++++++++++++++++++++++----------------
 1 files changed, 73 insertions(+), 42 deletions(-)

diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index 09cba27..97309d4 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -1149,35 +1149,49 @@ 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;
-	int size, __size, i, pc;
+	struct trace_entry *ent;
+	int size, __size, i, pc, __cpu;
 	unsigned long irq_flags;
+	char *raw_data;
 
-	local_save_flags(irq_flags);
 	pc = preempt_count();
-
 	__size = SIZEOF_KPROBE_TRACE_ENTRY(tp->nr_args);
 	size = ALIGN(__size + sizeof(u32), sizeof(u64));
 	size -= sizeof(u32);
+	if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
+		     "profile buffer not large enough"))
+		return 0;
 
-	do {
-		char raw_data[size];
-		struct trace_entry *ent;
-		/*
-		 * Zero dead bytes from alignment to avoid stack leak
-		 * to userspace
-		 */
-		*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
-		entry = (struct kprobe_trace_entry *)raw_data;
-		ent = &entry->ent;
-
-		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);
-	} while (0);
+	/*
+	 * Protect the non nmi buffer
+	 * This also protects the rcu read side
+	 */
+	local_irq_save(irq_flags);
+	__cpu = smp_processor_id();
+
+	if (in_nmi())
+		raw_data = rcu_dereference(trace_profile_buf_nmi);
+	else
+		raw_data = rcu_dereference(trace_profile_buf);
+
+	if (!raw_data)
+		goto end;
+
+	raw_data = per_cpu_ptr(raw_data, __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;
+
+	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:
+	local_irq_restore(irq_flags);
 	return 0;
 }
 
@@ -1188,33 +1202,50 @@ 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;
-	int size, __size, i, pc;
+	struct trace_entry *ent;
+	int size, __size, i, pc, __cpu;
 	unsigned long irq_flags;
+	char *raw_data;
 
-	local_save_flags(irq_flags);
 	pc = preempt_count();
-
 	__size = SIZEOF_KRETPROBE_TRACE_ENTRY(tp->nr_args);
 	size = ALIGN(__size + sizeof(u32), sizeof(u64));
 	size -= sizeof(u32);
+	if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
+		     "profile buffer not large enough"))
+		return 0;
+
+	/*
+	 * Protect the non nmi buffer
+	 * This also protects the rcu read side
+	 */
+	local_irq_save(irq_flags);
+	__cpu = smp_processor_id();
+
+	if (in_nmi())
+		raw_data = rcu_dereference(trace_profile_buf_nmi);
+	else
+		raw_data = rcu_dereference(trace_profile_buf);
+
+	if (!raw_data)
+		goto end;
+
+	raw_data = per_cpu_ptr(raw_data, __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;
 
-	do {
-		char raw_data[size];
-		struct trace_entry *ent;
-
-		*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
-		entry = (struct kretprobe_trace_entry *)raw_data;
-		ent = &entry->ent;
-
-		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);
-	} while (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:
+	local_irq_restore(irq_flags);
 	return 0;
 }
 


-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com

[PATCH tracing/kprobes 2/3] x86: Add VIA processor instructions

[Masami Hiramatsu]

Add VIA processor's Padlock instructions(MONTMUL, XSHA1, XSHA256).

Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Reported-by: Ingo Molnar <mingo@elte.hu>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Frank Ch. Eigler <fche@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jason Baron <jbaron@redhat.com>
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
---

 arch/x86/lib/x86-opcode-map.txt |    8 +++++++-
 1 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/arch/x86/lib/x86-opcode-map.txt b/arch/x86/lib/x86-opcode-map.txt
index 59e20d5..78a0daf 100644
--- a/arch/x86/lib/x86-opcode-map.txt
+++ b/arch/x86/lib/x86-opcode-map.txt
@@ -469,7 +469,7 @@ a2: CPUID
 a3: BT Ev,Gv
 a4: SHLD Ev,Gv,Ib
 a5: SHLD Ev,Gv,CL
-a6:
+a6: GrpPDLK
 a7: GrpRNG
 a8: PUSH GS (d64)
 a9: POP GS (d64)
@@ -803,6 +803,12 @@ GrpTable: Grp16
 3: prefetch T2
 EndTable
 
+GrpTable: GrpPDLK
+0: MONTMUL
+1: XSHA1
+2: XSHA2
+EndTable
+
 GrpTable: GrpRNG
 0: xstore-rng
 1: xcrypt-ecb


-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com

[PATCH tracing/kprobes 0/3] tracing/kprobes: kprobe-tracer updates

[Masami Hiramatsu]

Hi Frederic,

Here is a few updates for kprobe-tracer which fix reported bugs and
is for catching up the latest ftrace.

 - Use dynamic percpu profiling buffer patch is for catching up
   the latest ftrace change.
 - Checking the result of create_event_dir() patch improves ftrace
   code pointed by Li Zefan (thanks!).
 - Add VIA instructions to x86 insn decoder patch fixes decoding
   bug reported by Ingo Molnar (thanks!).

Soon after this series, I'll send perf-kprobe patch series.

Thank you,

---

Masami Hiramatsu (3):
      tracing/ftrace: Fix to check create_event_dir() when adding new events
      x86: Add VIA processor instructions
      tracing/kprobes: Use global event profile buffer in kprobe tracer


 arch/x86/lib/x86-opcode-map.txt |    8 ++-
 kernel/trace/trace_events.c     |   29 +++++++---
 kernel/trace/trace_kprobe.c     |  115 +++++++++++++++++++++++++--------------
 3 files changed, 101 insertions(+), 51 deletions(-)

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com

[PATCH tracing/kprobes 3/3] tracing/ftrace: Fix to check create_event_dir() when adding new events

[Masami Hiramatsu]

Check result of create_event_dir() and add ftrace_event_call to
ftrace_events list only if it is succeeded. Thanks Li for pointing it out.

Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Frank Ch. Eigler <fche@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Jason Baron <jbaron@redhat.com>
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
---

 kernel/trace/trace_events.c |   29 +++++++++++++++++++++--------
 1 files changed, 21 insertions(+), 8 deletions(-)

diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index a4b7c9a..f03cda3 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -957,12 +957,14 @@ static int __trace_add_event_call(struct ftrace_event_call *call)
 	if (!d_events)
 		return -ENOENT;
 
-	list_add(&call->list, &ftrace_events);
 	ret = event_create_dir(call, d_events, &ftrace_event_id_fops,
 				&ftrace_enable_fops, &ftrace_event_filter_fops,
 				&ftrace_event_format_fops);
 	if (ret < 0)
-		list_del(&call->list);
+		pr_warning("Could not create directory of trace events/%s\n",
+			   call->name);
+	else
+		list_add(&call->list, &ftrace_events);
 	return ret;
 }
 
@@ -1124,10 +1126,15 @@ static void trace_module_add_events(struct module *mod)
 				return;
 		}
 		call->mod = mod;
+		ret = event_create_dir(call, d_events,
+				       &file_ops->id, &file_ops->enable,
+				       &file_ops->filter, &file_ops->format);
+		if (ret < 0) {
+			pr_warning("Could not create directory of trace "
+				   "point events/%s\n", call->name);
+			continue;
+		}
 		list_add(&call->list, &ftrace_events);
-		event_create_dir(call, d_events,
-				 &file_ops->id, &file_ops->enable,
-				 &file_ops->filter, &file_ops->format);
 	}
 }
 
@@ -1267,10 +1274,16 @@ static __init int event_trace_init(void)
 				continue;
 			}
 		}
+		ret = event_create_dir(call, d_events, &ftrace_event_id_fops,
+				       &ftrace_enable_fops,
+				       &ftrace_event_filter_fops,
+				       &ftrace_event_format_fops);
+		if (ret < 0) {
+			pr_warning("Could not create directory of trace "
+				   "point events/%s\n", call->name);
+			continue;
+		}
 		list_add(&call->list, &ftrace_events);
-		event_create_dir(call, d_events, &ftrace_event_id_fops,
-				 &ftrace_enable_fops, &ftrace_event_filter_fops,
-				 &ftrace_event_format_fops);
 	}
 
 	while (true) {


-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com

Re: [PATCH tracing/kprobes 0/3] tracing/kprobes: kprobe-tracer updates

[Steven Rostedt]

On Fri, 2009-09-25 at 11:19 -0700, Masami Hiramatsu wrote:
> Hi Frederic,
> 
> Here is a few updates for kprobe-tracer which fix reported bugs and
> is for catching up the latest ftrace.
> 
>  - Use dynamic percpu profiling buffer patch is for catching up
>    the latest ftrace change.
>  - Checking the result of create_event_dir() patch improves ftrace
>    code pointed by Li Zefan (thanks!).
>  - Add VIA instructions to x86 insn decoder patch fixes decoding
>    bug reported by Ingo Molnar (thanks!).
> 
> Soon after this series, I'll send perf-kprobe patch series.

Acked-by: Steven Rostedt <rostedt@goodmis.org>

-- Steve

Re: [PATCH tracing/kprobes 3/3] tracing/ftrace: Fix to check create_event_dir() when adding new events

[Frederic Weisbecker]

On Fri, Sep 25, 2009 at 11:20:54AM -0700, Masami Hiramatsu wrote:
> Check result of create_event_dir() and add ftrace_event_call to
> ftrace_events list only if it is succeeded. Thanks Li for pointing it out.
> 
> Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Cc: Jim Keniston <jkenisto@us.ibm.com>
> Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
> Cc: Andi Kleen <ak@linux.intel.com>
> Cc: Christoph Hellwig <hch@infradead.org>
> Cc: Frank Ch. Eigler <fche@redhat.com>
> Cc: Frederic Weisbecker <fweisbec@gmail.com>
> Cc: H. Peter Anvin <hpa@zytor.com>
> Cc: Ingo Molnar <mingo@elte.hu>
> Cc: Jason Baron <jbaron@redhat.com>
> Cc: K.Prasad <prasad@linux.vnet.ibm.com>
> Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
> Cc: Li Zefan <lizf@cn.fujitsu.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
> Cc: Tom Zanussi <tzanussi@gmail.com>
> ---
> 
>  kernel/trace/trace_events.c |   29 +++++++++++++++++++++--------
>  1 files changed, 21 insertions(+), 8 deletions(-)
> 
> diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
> index a4b7c9a..f03cda3 100644
> --- a/kernel/trace/trace_events.c
> +++ b/kernel/trace/trace_events.c
> @@ -957,12 +957,14 @@ static int __trace_add_event_call(struct ftrace_event_call *call)
>  	if (!d_events)
>  		return -ENOENT;
>  
> -	list_add(&call->list, &ftrace_events);
>  	ret = event_create_dir(call, d_events, &ftrace_event_id_fops,
>  				&ftrace_enable_fops, &ftrace_event_filter_fops,
>  				&ftrace_event_format_fops);
>  	if (ret < 0)
> -		list_del(&call->list);
> +		pr_warning("Could not create directory of trace events/%s\n",
> +			   call->name);




We already have such warnings in event_create_dir, with even more granularity
against the failure reason. The patch is fine, but I'll just remove the warning
introduced inside while applying it.

Thanks.


PS: Also I guess we could add the event in the list right from
event_create_dir()...

Re: [PATCH tracing/kprobes 0/3] tracing/kprobes: kprobe-tracer updates

[Frederic Weisbecker]

On Fri, Sep 25, 2009 at 11:19:52AM -0700, Masami Hiramatsu wrote:
> Hi Frederic,
> 
> Here is a few updates for kprobe-tracer which fix reported bugs and
> is for catching up the latest ftrace.
> 
>  - Use dynamic percpu profiling buffer patch is for catching up
>    the latest ftrace change.
>  - Checking the result of create_event_dir() patch improves ftrace
>    code pointed by Li Zefan (thanks!).
>  - Add VIA instructions to x86 insn decoder patch fixes decoding
>    bug reported by Ingo Molnar (thanks!).
> 
> Soon after this series, I'll send perf-kprobe patch series.
> 
> Thank you,


Applied, thanks Masami!