PR#756 fixed the data corruption bug, but didn't clear `ip`. PR#760
fixed that off-by-one error, but missed the case where `ip == filledIp`,
which is harder for the fuzzers to find (it took 20 days not 1 day).
Verified this fixed the issue reported by OSS-Fuzz.
Credit to OSS-Fuzz.
ip -= matchCode - newMatchCode;
assert(newMatchCode < matchCode);
matchCode = newMatchCode;
- if (unlikely(ip < filledIp)) {
+ if (unlikely(ip <= filledIp)) {
/* We have already filled up to filledIp so if ip ends up less than filledIp
* we have positions in the hash table beyond the current position. This is
* a problem if we reuse the hash table. So we have to remove these positions