[FFmpeg-devel] [PATCH] random_seed: Limit the time taken by get_generic_seed

[Martin Storsjö]

On Fri, 31 Jan 2025, Michael Niedermayer wrote:

>> diff --git a/libavutil/random_seed.c b/libavutil/random_seed.c
>> index 8a4e4f1fc0..8f969060a0 100644
>> --- a/libavutil/random_seed.c
>> +++ b/libavutil/random_seed.c
>> @@ -83,6 +83,7 @@ static uint32_t get_generic_seed(void)
>>      static uint32_t buffer[512] = { 0 };
>>      unsigned char digest[20];
>>      uint64_t last_i = i;
>> +    int cur_iters = 0;
>>
>>      av_assert0(sizeof(tmp) >= av_sha_size);
>>
>> @@ -98,11 +99,13 @@ static uint32_t get_generic_seed(void)
>>
>>      for (;;) {
>>          clock_t t = clock();
>> -        if (last_t + 2*last_td + (CLOCKS_PER_SEC > 1000) >= t) {
>> +        if (last_t + 2*last_td + (CLOCKS_PER_SEC > 1000) >= t && cur_iters < 128) {
>>              last_td = t - last_t;
>>              buffer[i & 511] = 1664525*buffer[i & 511] + 1013904223 + (last_td % 3294638521U);
>> +            cur_iters++;
>>          } else {
>>              last_td = t - last_t;
>> +            cur_iters = 0;
>
> Iam concerned this could negatively impact entropy
> The "else" should be run when a interrupt/task switch occured.
> If that doesnt occur in 128 iterations that doesnt gurantee the entropy
> has increased.
>
> If there are only 0 and 1, ideally we should look at the distribution and
> go to the else when the pattern differs from the past / has some signs of
> randomness

Ok, I've somewhat tried to implement this, please take a look.

// Martin