DavidKorczynski
DavidKorczynski
this includes calltree generation and fuzz blockers
The table of contents gets a bit of a messy look when there are many fuzzers. It would be nice to structure it a bit nicer, perhaps make it collapsible...
LTO can be incompatible with certain projects, in particular larger projects. Both in terms of resource exhaustion and pure compatibility (i.e. breaking builds). An option is to enable a mode...
We currently have to build the plugin as part of clang rather than an out-of-tree plugin. This problem is described here in the code: https://github.com/ossf/fuzz-introspector/blob/9c0ea223b5148b71bc774842404a2317ba70a778/llvm/lib/Transforms/FuzzIntrospector/FuzzIntrospector.cpp#L1002-L1005 Specifically, we run into this...
Currently the calltree displays only a single coloring of nodes, which is based off a per target coverage report in most cases. However, in general, the calltree can be used...
We should add docs that go into details with the analyses and data structures in fuzz-introspector. Down the line it's likely a lot more analyses (plugins) will be made and...
It would be nice to be able to fold and unfold all functions by depth. This makes it much easier to quickly inspect all functions within a given function.
Umbrella issue for projects failing. - bitcoin-core [issue](https://github.com/ossf/fuzz-introspector/issues/27) - nss. - ~~Compilation problem~~ [problem identified and fix is coming up](https://github.com/ossf/fuzz-introspector/issues/30#issuecomment-1010277706). The problem was resource exhaustion, we simply need to limit...
One of the main things I have in mind here is that it would be nice to re-run some of the fuzz-introspector analysis by way of the interface, without having...
When we use the fuzz-introspector on projects with a lot of functions, e.g. OpenSSL, then the resulting HTML report is huge. We should refactor the HTML report so this won't...