#32 Integer underflow in decodeInteger32/64

[Siebren W]

Since the 0.9.6's newest range check, numbers with the value -1 of type i64 or i32 will result in -107 (EXI_ERROR_INTEGER_OVERFLOW)

Cause: the raw result of decodeUnsignedInteger32 gets a substraction before comparing it for overflow, making the number wrap.

Solution:

remove the -1 from `if ((uint32-1) < 2147483647) {' (same for the uint64)

int16 does not seem to be affected.

[Sebastian Bode]

Fixed in revision r127 and released in OpenV2G 0.9.7.

The unsigned subtraction in the original range check caused decoding of −1 to fail. The correct check is uint32 <= INT32_MAX (and analogously for 64-bit values), so that the minimum signed value is also accepted. The signed-value reconstruction was adjusted to avoid overflow.

Thanks for reporting this issue!

Release: https://sourceforge.net/projects/openv2g/files/release/OpenV2G_0.9.7/