Summary: Consuming bytes from the end of data instead of from the front to prevent "all-in-one" decisions.
Test Plan:
Reviewers:
Subscribers:
Tasks:
Tags:
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.
When using an empty dictionary, we bail out of loading or attaching it in
ways that leave the working context in potentially slightly different states.
In particular, in some paths, we will cause the currentOffset to be non-zero,
while in others we would allow it to remain 0.
This difference in behavior is perfectly harmless, but in some situations, it
can produce slight differences in the compressed output. For sanity's sake,
we currently try to maintain a strict correspondence between the behavior of
the dict attachment and the dict loading paths. This patch restores them to
behaving identically.
This shouldn't have any negative side-effects, as far as I can tell. When
writing the dict attachment code, I tried to preserve zeroed currentOffsets
when possible, since they benchmarked as very slightly faster. However, the
case of attaching an empty dictionary is probably rare enought that it's
acceptable to minisculely degrade performance in that corner case.
When the match is very long and found quickly, we can do
matchLength * nbCompares iterations through the chain
swapping, which can really slow down compression.
It is important to continue to look backwards if the current pattern
reaches `lowPrefixPtr`. If the pattern detection doesn't go all the
way to the beginning of the pattern, or the end of the pattern it
slows down the search instead of speeding it up.
The slow unit in `round_trip_stream_fuzzer` used to take 12 seconds
to run with -O3, now it takes 0.2 seconds.
Credit to OSS-Fuzz
This diff fixes an issue in which we failed to clear the `dictCtx` in HC
compression. The `dictCtx` is not supposed to be used when an `extDict` is
present: matches found in the `dictCtx` do not account for the presence of an
`extDict` segment, and their offsets are therefore miscalculated when one is
present. This can lead to data corruption.
This diff clears the `dictCtx` whenever setting an `extDict`.
This issue was uncovered by @terrelln's fuzzing work.
