The pyasn1 library's univ.Real type converts (mantissa, base, exponent) tuples from decoded ASN.1 REAL values into Python floats using exact big-integer exponentiation. When the exponent is very large, this conversion attempts to create an astronomically large integer, causing excessive CPU and memory consumption. This resource exhaustion can hang the process during any float conversion operation such as prettyPrint(), str(), comparisons, arithmetic, or explicit float() calls. The vulnerability affects pyasn1.type.univ.Real objects produced by BER, CER, and DER decoders from untrusted input. The issue does not occur during decoding alone but only when float conversion is triggered. The vulnerability is fixed in pyasn1 version 0.6.4, which uses math.ldexp() for binary values and raises OverflowError for decimal values with out-of-range exponents, preventing large intermediate integer construction.

This vulnerability allows an attacker to cause denial of service by supplying ASN.1 data with REAL values that have very large exponents. When the application converts these decoded REAL values to floats (via printing, logging, comparison, or arithmetic), it consumes excessive CPU and memory, potentially hanging or crashing the process. There is no impact on confidentiality or integrity, only availability is affected.

A fix is available in pyasn1 version 0.6.4. Users should upgrade to this version to prevent the vulnerability. Until upgraded, avoid converting, printing, or comparing decoded Real objects from untrusted sources. Instead, inspect the raw (mantissa, base, exponent) tuple to avoid triggering float conversion.