[llvm-dev] RFC: Strong GC References in LLVM

[Sanjoy Das via llvm-dev]

Hi Daniel,

Daniel Berlin wrote:
 >     // ptr_a and ptr_b are NoAlias, ptr_a holds 0 to begin with.
 >
 >     ThreadA:
 >        while(true) { store volatile i32 42, i32* %ptr_b }
 >        store atomic i32 42, i32* %ptr_a
 >
 >     ThreadB:
 >        %val = load atomic i32, i32* %ptr_a
 >        assert(%val is not 42)  // The store is "guarded" by an inf loop
 >
 >
 >     We can't reorder the store to ptr_a to before the infinite loop.  The
 >     volatile store is there to make the infinite loop well defined.
 >
 >
 > These do not have hidden control flow. It is actually well defined it
 > just literallly involves other instructions :)

As written above, sure; but the `while (true)` could be inside a
nounwind function (that does not return).  So it is not sufficient to
just look for calls to unknown functions in a function body to
conclude that it is alwaysreturn, it isn't sufficient to look for
llvm::Loop's either since we could have cycles not describable as
loops.

I'm not claiming that ^ is new information, btw, I'm justifying why I
specifically brought up the "while(true)" example. :)

 > Note that gcc will optionally connect the infinite loop itself to the
 > exit block with a fake edge if you want
 > (you can add/remove fake edges on a per-opt basis).

Does GCC's notion of a "loop" (unlike LLVM) include all potentially
infinite control flow?

-- Sanjoy