fuzz-introspector
fuzz-introspector copied to clipboard
ossf
Fuzz Introspector -- introspect, extend and optimise fuzzers
Where `numpy` is imported into a `fuzz.py` file it's getting included in the reachability and coverage numbers A couple of examples - [bottle - introspector report - 2023-05-21](https://storage.googleapis.com/oss-fuzz-introspector/bottleneck/inspector-report/20230521/fuzz_report.html) - [toml...
Looking at a recent croniter coverage report [https://storage.googleapis.com/oss-fuzz-introspector/croniter/inspector-report/20230501/fuzz_report.html](https://storage.googleapis.com/oss-fuzz-introspector/croniter/inspector-report/20230501/fuzz_report.html) the "Functions Statically Reachable" is very low `2/48`. However function coverage is pretty good. It looks like there are lots of functions...
I recently integrated the [cachetools](https://github.com/tkem/cachetools) library into OSS-Fuzz recently. This library uses a `@cached` decorator to indicate a function output to cache, e.g. ``` python @cachetools.cached(cache=fdp.PickValueInList(CACHE_TYPES), lock=cache_lock, info=fdp.ConsumeBool()) def fib(n):...
Using "is reached" to mean "is statically reachable" is confusing. Maybe say "is statically reachable" or "is reachable"
I found a problem related to the hyperlink of blocked branch in OpenSSL report. This could lead to two pieces of mis-leading information, wrong link to the coverage report and...
For example, see the call site link of the function `igraph_i_parse_real` in the igraph report, https://storage.googleapis.com/oss-fuzz-introspector/igraph/inspector-report/20230226/calltree_view_4.html?scrollToNode=00221 The call site of the aforementioned function in the original igraph library is this:...
@eantelman just following up on this here: *Just going by the name of this ticket, this is another odd behavior, less interesting than how to handle intermediate generated source files...
https://oss-fuzz-introspector.storage.googleapis.com/index.html recently got a facelift, and this made some issues obvious. One is that code coverage is often a lot higher than reachability for Python projects. We should investigate why...
This will make tools such as auto-fuzz https://github.com/ossf/fuzz-introspector/tree/main/tools/auto-fuzz much nicer.
Migrated from https://github.com/ossf/fuzz-introspector/pull/680#issuecomment-1347330552 The `angus-mail` java project is missing data when analysed by the jvm frontend. For example the calltree for the BASE64EncoderStreamFuzzer we get is: ``` Call tree fuzzerTestOneInput(com.code_intelligence.jazzer.api.FuzzedDataProvider)...