[clang] Clang ssaf driver 4 (PR #231999)

[Aaron Ballman via cfe-commits]

AaronBallman wrote:

> @AaronBallman FYI for suspicious committer activity. These two people landed a lot of really big changes (starting at #231583, ending at this PR) without upstream review (time between PR submit and approval usually 2 minutes).

Thank you for bringing this to my attention!

> I can only assume that this a code dump that was reviewed downstream in advance, but I don't believe that such code dumps are generally considered acceptable, at least without prior coordination. I'm not sure if such coordination happened somewhere, so putting this on your radar.
> 
> If these were commited in LLVM, I'd be reverting all of these PRs about now.

Agreed, these should be reverted so they can be properly reviewed by the community. I also noticed that the RFCs for SSAF were never actually accepted, making the way these were landed especially concerning: 

https://discourse.llvm.org/t/rfc-scalable-static-analysis-framework/88678
https://discourse.llvm.org/t/rfc-ssaf-entity-linker/89593

(Looking through the RFC comments, I don't see any opposition to them and so I don't think there's anything malicious happening here, but pushing changes with no opportunity for community review on functionality that was never officially accepted through the RFC process has really bad optics. I'll talk to others in the Clang Area Team about what we can do to close the loop on those.)

> these changes violate several layering principles, e.g. making GNU Make a mandatory dependency for lit tests and introducing dependency from clang on clang-tools-extra and libc++.

Even if there was some prior coordination, this kind of change requires an RFC.

> Given that they are in PST, I'll take this over from them and push a revert of the full stack.

Thank you for the help!

https://github.com/llvm/llvm-project/pull/231999