NagyDonat wrote: The CI run validated that this PR does not introduce any new crash and any result change on a set of open source projects (memcached, tmux, curl, twin, vim, openssl, sqlite, ffmpeg, postgres, tiny-json, inih, argparse, linenoise, tinyxml2, libwebm, xerces, bitcoin, protobuf, qtbase, contour, acid, openrct2).
I also reviewed the old commit https://github.com/llvm/llvm-project/commit/d65379c8d4768ddae143672f5d4827acafe22553 and confirmed that it only extends some `ProgramPoint` subclasses with an extra data member (a `ConstCFGElementRef`) to distinguish some `ProgramPoint`s that were previously identical (and caused an unwanted loop in the exploded graph). This would _very_ slightly increase the size of the exploded graph by preventing logically questionable deduplication, but I don't expect a measurable performance hit. I can also confirm that there is no code smell (apart from touching preexisting smelly code, which led to the creation of https://github.com/llvm/llvm-project/issues/233598), so I'm now confident that fixing and "unleashing" this old commit is not scary. https://github.com/llvm/llvm-project/pull/231515 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
