[libvirt] [PATCH 1/2] util: Don't overflow in virRandomBits

[Eric Blake]

On 08/01/2018 07:16 AM, Daniel P. Berrangé wrote:
> On Wed, Aug 01, 2018 at 01:44:32PM +0200, Michal Privoznik wrote:
>> The function is supposed to return up to 64bit long integer. In
>> order to do that it calls virRandomBytes() to fill the integer
>> with random bytes and then masks out everything but requested
>> bits. However, when doing that it shifts 1U and not 1ULL. So
>> effectively, requesting 32 random bis or more always return 0
>> which is not random enough.
>>
>> Signed-off-by: Michal Privoznik <mprivozn at redhat.com>
>> ---
>>   src/util/virrandom.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/src/util/virrandom.c b/src/util/virrandom.c
>> index 01cc82a052..3c011a8615 100644
>> --- a/src/util/virrandom.c
>> +++ b/src/util/virrandom.c
>> @@ -68,7 +68,7 @@ uint64_t virRandomBits(int nbits)
>>           return 0;
>>       }
>>   
>> -    ret &= (1U << nbits) - 1;
>> +    ret &= (1ULL << nbits) - 1;

1ULL << 64 is undefined in C. We need to write this as:

if (nbits < 64)
     ret &= (1ULL << nbits) - 1;


-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org