In IGMPv2, the Max response time is an 8-bit value each unit encoding 0.1 seconds. The typical max response time is 100 resulting in a 10 second max response time.

In IGMPv3, if the Max response code is < 128, the calculated time is the same. But per https://tools.ietf.org/html/rfc3376 section 4.1.1:

   If Max Resp Code < 128, Max Resp Time = Max Resp Code

   If Max Resp Code >= 128, Max Resp Code represents a floating-point
   value as follows:

       0 1 2 3 4 5 6 7
      |1| exp | mant  |

   Max Resp Time = (mant | 0x10) << (exp + 3)

https://tools.ietf.org/html/rfc6636 section 4.2 gives an example:

For example, if one wants to set the Max
   Response Time to 20.0 seconds, the Max Resp Code should be expressed
   as "0b10001001", which is divided into "mant=0b1001" and "exp=0b000".

I don't see how 0b10001001 results in 20 seconds. If the mantissa is 0b1001 (9 decimal) and we bitshift left 3 (exponent 0 + 3) that results in 0b01001000 = 72 decimal or 7.2 seconds.

How do I calculate the IGMPv3 response time for max response code values >= 128?

Neither RFC has relevant errata.


I think the part you misunderstood in RFC 3376 Section 4.1.1 that you quoted:

Max Resp Time = (mant | 0x10) << (exp + 3)

I think it could be clearer (not mixing binary and hexadecimal), but as far as I can gather:

(1001 OR 10000) = 11001 -> 11001^11 = 11001000 = Decimal 200

You must OR 0x10 to the matissa, which is one binary digit larger than the mantissa, so you prepend a binary 1 to the mantissa before shifting the exponent + 3.

This is something that perhaps should be sent in as errata.


Ron's hint is very close - in binary floats, the leading 1 is implied and not actually stored in the representation. It simply doesn't make sense to have anything else but a leading 1 in the (normalized) mantissa, so actually storing it would waste a (precision) bit.

The | 0x10 part in the formula expresses this.

See https://en.wikipedia.org/wiki/Floating-point_arithmetic#Internal_representation for the IEEE 754 format:

In the IEEE binary interchange formats the leading 1 bit of a normalized significand is not actually stored in the computer datum. It is called the "hidden" or "implicit" bit.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.