Re: [PATCH V6 05/12] perf tools: tidy up sample parsing overflowchecking

[Adrian Hunter]

On 16/07/13 14:53, Jiri Olsa wrote:
> On Tue, Jul 16, 2013 at 09:38:11AM +0300, Adrian Hunter wrote:
>> The size of data retrieved from a sample event must be
>> validated to ensure it does not go past the end of the
>> event.  That was being done sporadically and without
>> considering integer overflows.
>>
>> Signed-off-by: Adrian Hunter <adrian.hunter@xxxxxxxxx>
>> ---
>>  tools/perf/util/evsel.c | 102 ++++++++++++++++++++++++++++++------------------
>>  1 file changed, 64 insertions(+), 38 deletions(-)
>>
>> diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
>> index 724b75a..20e2ed9 100644
>> --- a/tools/perf/util/evsel.c
>> +++ b/tools/perf/util/evsel.c
>> @@ -1114,24 +1114,38 @@ static int perf_evsel__parse_id_sample(const struct perf_evsel *evsel,
>>  	return 0;
>>  }
>>  
>> -static bool sample_overlap(const union perf_event *event,
>> -			   const void *offset, u64 size)
>> +static inline bool overflow_one(const void *endp, const void *offset)
>>  {
>> -	const void *base = event;
>> -
>> -	if (offset + size > base + event->header.size)
>> -		return true;
>> +	return offset + sizeof(u64) > endp;
>> +}
>>  
>> -	return false;
>> +static inline bool overflow(const void *endp, u16 max_size, const void *offset,
>> +			    u64 size)
>> +{
>> +	return size > max_size || offset + size > endp;
> 
> hum, does '(size > max_size)' catch anything that would
> not be catched by '(offset + size > endp)' ?

'offset + size' can overflow and end up less than endp

> 
>>  }
>>  
>> +#define OVERFLOW_CHECK_ONE(offset)				\
>> +	do {							\
>> +		if (overflow_one(endp, (offset)))		\
>> +			return -EFAULT;				\
>> +	} while (0)
>> +
>> +#define OVERFLOW_CHECK(offset, size)				\
>> +	do {							\
>> +		if (overflow(endp, max_size, (offset), (size)))	\
>> +			return -EFAULT;				\
>> +	} while (0)
> 
> I think, it'd be better to have just one overflow check function, like:
> 
> static inline bool overflow(const void *endp, const void *offset, u64 size)
> {
>         return offset + size > endp;
> }
> 
> #define OVERFLOW_CHECK(offset, size)                            \
>         do {                                                    \
>                 if (overflow(endp, (offset), (size)))           \
>                         return -EFAULT;                         \
>         } while (0)
> 
> #define OVERFLOW_CHECK_u64(offset) OVERFLOW_CHECK(offset, sizeof(u64))
> 

OK

>> +
>>  int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event,
>>  			     struct perf_sample *data)
>>  {
> 
> Hum, I was thinking of making this check more general and
> cover also perf_event__synthesize_sample function.. but
> looks like it's not needed there..?

Currently the caller must ensure there is enough space.

> 
> 
>>  	u64 type = evsel->attr.sample_type;
>> -	u64 regs_user = evsel->attr.sample_regs_user;
>>  	bool swapped = evsel->needs_swap;
>>  	const u64 *array;
>> +	u16 max_size = event->header.size;
>> +	const void *endp = (void *)event + max_size;
>> +	u64 sz;
>>  
>>  	/*
>>  	 * used for cross-endian analysis. See git commit 65014ab3
>> @@ -1153,6 +1167,11 @@ int perf_evsel__parse_sample(struct perf_evsel *evsel, union perf_event *event,
>>  
>>  	array = event->sample.array;
>>  
>> +	/*
>> +	 * sample_size is based on PERF_SAMPLE_MASK which includes up to
> 
> please prepend ^^^: The evsel's

OK

> 
>> +	 * PERF_SAMPLE_PERIOD.  After that overflow_one() or overflow() must be
>> +	 * used to check the format does not go past the end of the event.
>> +	 */
>>  	if (evsel->sample_size + sizeof(event->header) > event->header.size)
>>  		return -EFAULT;
>>  

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/