[PATCH] D138958: [clang] Better UX for Clang’s unwind-affecting attributes

[Roman Lebedev via Phabricator via cfe-commits]

lebedev.ri marked an inline comment as done.
lebedev.ri added inline comments.


================
Comment at: clang/lib/Analysis/CFG.cpp:2725
       NoReturn = true;
-    if (FD->hasAttr<NoThrowAttr>())
+    if (FD->hasAttr<NoUnwindAttr>() || FD->hasAttr<NoThrowAttr>())
       AddEHEdge = false;
----------------
erichkeane wrote:
> lebedev.ri wrote:
> > erichkeane wrote:
> > > I find myself thinking we should probably have a function in FunctionDecl that tests for the various states of the function, rather than keep checking the attribute's presence.  
> > Yes, we have a huge spaghetti code spread through clang
> > that tries to answer the same two question:
> > 1. i'm a caller, if i call this function, might it throw?
> > 1. i'm a callee, what should i do with exceptions that try to unwind out of me?
> > 
> > I don't know how to improve that, and i don't think just moving this into FunctionDecl would help.
> > 
> `FunctionDecl::WontThrow` and `FunctionDecl::ShouldUnwind` ? 
As noted in the RFC, there are 3 possible behaviors on unwind.
If we want to improve interface, we should account for all of them.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D138958/new/

https://reviews.llvm.org/D138958