[2/2] PM: Fix Oops from NULL pointer dereference in wakeup_source_activate

Commit Message

Steven Rostedt Nov. 21, 2013, 2:26 a.m. UTC

On Wed, 20 Nov 2013 19:01:22 -0700
Shuah Khan <[e-mail]> wrote:

> On 11/20/2013 06:40 PM, Shuah Khan wrote:

> > ---
> >   include/trace/events/power.h | 6 ++++--
> >   1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/include/trace/events/power.h b/include/trace/events/power.h
> > index cda100d..5ba545a 100644
> > --- a/include/trace/events/power.h
> > +++ b/include/trace/events/power.h
> > @@ -110,12 +110,14 @@ DECLARE_EVENT_CLASS(wakeup_source,
> >   	TP_ARGS(name, state),
> >
> >   	TP_STRUCT__entry(
> > -		__string(       name,           name            )
> > +		__string(name,  name ? name : "(no name)")
> >   		__field(        u64,            state           )
> >   	),
> >
> >   	TP_fast_assign(
> > -		__assign_str(name, name);
> > +		const char *tname = name ? name : "(no name)";
> > +
> > +		__assign_str(name, tname);
> >   		__entry->state = state;
> >   	),
> >
> >
> 
> Adding tracing maintainers.

Thanks!

This is one solution, but what about just making the tracing facility a
bit more robust for everyone. Following what glibc printf() does when
it is passed a NULL, does this patch fix it too?

-- Steve

(haven't even compile tested this)


--
To unsubscribe from this list: send the line "unsubscribe linux-pm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Shuah Khan Nov. 21, 2013, 6:02 p.m. UTC | #1

On 11/20/2013 07:26 PM, Steven Rostedt wrote:
> On Wed, 20 Nov 2013 19:01:22 -0700
> Shuah Khan <[e-mail]> wrote:
>
>> On 11/20/2013 06:40 PM, Shuah Khan wrote:
>
>>> ---
>>>    include/trace/events/power.h | 6 ++++--
>>>    1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/include/trace/events/power.h b/include/trace/events/power.h
>>> index cda100d..5ba545a 100644
>>> --- a/include/trace/events/power.h
>>> +++ b/include/trace/events/power.h
>>> @@ -110,12 +110,14 @@ DECLARE_EVENT_CLASS(wakeup_source,
>>>    	TP_ARGS(name, state),
>>>
>>>    	TP_STRUCT__entry(
>>> -		__string(       name,           name            )
>>> +		__string(name,  name ? name : "(no name)")
>>>    		__field(        u64,            state           )
>>>    	),
>>>
>>>    	TP_fast_assign(
>>> -		__assign_str(name, name);
>>> +		const char *tname = name ? name : "(no name)";
>>> +
>>> +		__assign_str(name, tname);
>>>    		__entry->state = state;
>>>    	),
>>>
>>>
>>
>> Adding tracing maintainers.
>
> Thanks!
>
> This is one solution, but what about just making the tracing facility a
> bit more robust for everyone. Following what glibc printf() does when
> it is passed a NULL, does this patch fix it too?
>

Why not? I am working to get the code compiling now. Will let you know 
if this works.

-- Shuah

Patch

diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h
index 52594b2..bdac88c 100644
--- a/include/trace/ftrace.h
+++ b/include/trace/ftrace.h
@@ -372,7 +372,8 @@  ftrace_define_fields_##call(struct ftrace_event_call *event_call)	\
 	__data_size += (len) * sizeof(type);
 
 #undef __string
-#define __string(item, src) __dynamic_array(char, item, strlen(src) + 1)
+#define __string(item, src) __dynamic_array(char, item,			\
+				    strlen((src) ? (src) : "(null)") + 1)
 
 #undef DECLARE_EVENT_CLASS
 #define DECLARE_EVENT_CLASS(call, proto, args, tstruct, assign, print)	\
@@ -501,7 +502,7 @@  static inline notrace int ftrace_get_offsets_##call(			\
 
 #undef __assign_str
 #define __assign_str(dst, src)						\
-	strcpy(__get_str(dst), src);
+	strcpy(__get_str(dst), (src) ? (src) : "(null)");
 
 #undef TP_fast_assign
 #define TP_fast_assign(args...) args