[LLVMdev] LLVM Inliner

[Chris Lattner]

On Nov 29, 2010, at 8:40 AM, Duncan Sands wrote:

>> 
>> 
>> So bottom up traversal of the (scc node of) callgraph is not sufficient to
>> guarantee all callees are processed before the caller?
> 
> that's not the problem, the problem is that the pass manager schedules two
> bottom up traversals, one for the inliner + cleanup passes and another for
> the function attributes pass, if you place place the function attributes
> pass later in the pass list.  I don't know why, but it seems to be a pass
> manager bug (perhaps it is a feature...).

I'm not sure what the issue is, but that is clearly a bug.

-Chris