[PATCH] D35994: Debug info for variables whos type is shrinked to bool

[David Blaikie via llvm-commits]

On Fri, Jul 28, 2017 at 8:31 AM Adrian Prantl via Phabricator via
llvm-commits <llvm-commits at lists.llvm.org> wrote:

> aprantl added inline comments.
>
>
> ================
> Comment at: lib/Transforms/IPO/GlobalOpt.cpp:1577
> +  for(auto *GV : GVs)
> +    NewGV->addDebugInfo(GV);
> +
> ----------------
> I'm not familiar with this transformation: Do we need to add a
> DIExpression to mask out all but the first bit (i.e. can multiple bools be
> packed into the same uint32_t such that it could confuse debuggers)?
>
>
> ================
> Comment at: test/Transforms/GlobalOpt/static-global-boolean-dwarf.c:2
> +// RUN: clang -g -O2 -c %s -o %t
> +// RUN: llvm-dwarfdump %t | FileCheck %s
> +
> ----------------
> We shouldn't require a frontend to be available when running tests. Can
> you use thew output of clang -S -emit-llvm instead an run it through opt?
>

Also I imagine the test could be further minimized? Is the enum necessary,
for example (oh, perhaps since it's C it is? not sure - I'd probably write
it in C++ using 'bool')? Could it be simplified by having the boolean value
(but not its address) leak out of the translation unit - though I guess
that's what 'get_foo' does. Does init_boolean need to be that complicated?
Or could it be : "void set_foo(bool b) { foo = b; }" & still trigger the
right amount of optimization? (so I'd expect the test case to come down to:
"static bool foo; void set_foo(bool b) { foo = b; } bool get_foo() { return
foo; }")


>
>
> https://reviews.llvm.org/D35994
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170728/e80dc587/attachment.html>