[llvm-commits] [llvm] r80215 - in /llvm/trunk: include/llvm/Transforms/IPO/InlinerPass.h lib/Transforms/IPO/Inliner.cpp test/Transforms/Inline/array_merge.ll

[Chris Lattner]

On Aug 27, 2009, at 1:34 AM, Evan Cheng wrote:

>
>
> On Aug 26, 2009, at 11:44 PM, Daniel Dunbar <daniel at zuster.org> wrote:
>
>> On Wed, Aug 26, 2009 at 11:29 PM, Chris Lattner<sabre at nondot.org>  
>> wrote:
>>> One piece of concern that I didn't look into: at -O0 -g with
>>> forced inlining this will almost certainly result in worse debug
>>> info.  I think this is acceptable though given that this is a case
>>> of "debugging optimized code", and we don't want debug info to
>>> prevent the optimizer from doing things anyway.
>>
>> Is there a good reason not to distinguish this situation
>> (optimize-without-hurting-debugging)? Yes, the user asked for
>> inlining, but at -O0 -g they probably still want good debug info. We
>> could easily pass this flag into the inliner.
>
> I agree. I'd rather this optimization is disabled at -O0. Is that  
> possible?

On further reflection, this xform shouldn't hurt debug info at all:  
two user variables can have the same location with no problem, and  
they have non-overlapping live ranges (by definition).  Note that -O0  
only inlines always inline functions anyways.

-Chris