[PATCH] D11969: Avoid using of DataLayout::getPointerSize() without address space argument in DWARF info generation

[Mekhanoshin, Stanislav via llvm-commits]

Thanks Matt! I have verified this test passes with my change.

Stas

-----Original Message-----
From: Matt Arsenault [mailto:whatmannerofburgeristhis at gmail.com] On Behalf Of Matt Arsenault
Sent: 12 августа 2015 г. 0:50
To: Mekhanoshin, Stanislav <Stanislav.Mekhanoshin at amd.com>
Cc: Justin Bogner <mail at justinbogner.com>; Stanislav Mekhanoshin via llvm-commits <llvm-commits at lists.llvm.org>; mehdi_amini at apple.com; reviews+D11969+public+c218119e32184906 at reviews.llvm.org
Subject: Re: [PATCH] D11969: Avoid using of DataLayout::getPointerSize() without address space argument in DWARF info generation


> On Aug 12, 2015, at 12:04 AM, Mekhanoshin, Stanislav via llvm-commits <llvm-commits at lists.llvm.org> wrote:
> 
> That's the problem with the test, I know it is generally needed. There are no such targets in the llvm with a data layout like this, and until data layout issue with pointer size is fixed it will hardly appear.

AMDGPU has been using this data layout for a long time


> AMDGPU could be the good candidate to write such a test for, but again it cannot afford to switch data layout until these issues are fixed.

The test in this patch can serve as this test since it should start producing the right result http://reviews.llvm.org/D11972