[llvm-dev] Let CallGraphSCCPass Use Function-Level Analysis

Mikhail Gudim via llvm-dev llvm-dev at lists.llvm.org
Thu Jan 9 07:35:48 PST 2020


Thanks Brian and Johannes for your replies.

    " As Brian noted, stuff like this works better in the new pass manager."

Yes, but I need it to work for the old pass manager.

	" Did you initialize the pass, via
     `INITIALIZE_PASS_DEPENDENCY(PostDominatorTreeWrapperPass)`?
     Did you require it, via
     ` AU.addRequired<PostDominatorTreeWrapperPass>();`"

Yes, I did.

    " Btw. May I ask what you are planning to do?"

I would like to iterate through all post-dominators of the entry block (instead of just the entry block) to find more opportunities for argument promotion:

https://reviews.llvm.org/D72382

Btw, it would be great if you could review / give some feedback on that patch.
-----Original Message-----
From: Doerfert, Johannes [mailto:jdoerfert at anl.gov] 
Sent: Tuesday, January 07, 2020 12:19 PM
To: Brian Gesiak <modocache at gmail.com>
Cc: Mikhail Gudim <mikhail.gudim at huawei.com>; llvm-dev at lists.llvm.org
Subject: Re: [llvm-dev] Let CallGraphSCCPass Use Function-Level Analysis

Hi Mikhail,

As Brian noted, stuff like this works better in the new pass manager.

Even in the old pass manager I thought it should work though.
Did you initialize the pass, via
 `INITIALIZE_PASS_DEPENDENCY(PostDominatorTreeWrapperPass)`?
Did you require it, via
` AU.addRequired<PostDominatorTreeWrapperPass>();`?

Btw. May I ask what you are planning to do?

Cheers,
  Johannes


On 01/07, Brian Gesiak via llvm-dev wrote:
> Hello! The new pass manager provides analysis proxies from one IR unit 
> type to another. These are specializations of 
> 'llvm::InnerAnalysisManagerProxy'. For example, 
> 'llvm::FunctionAnalysisManagerModuleProxy' allows you to access 
> function analyses from within module passes.
> 
> In your case, it sounds like you'd want to use 
> 'llvm::FunctionAnalysisManagerCGSCCProxy', which allows you to access 
> function analyses from within an CGSCC pass. Here's an example of 
> using it:
> 
> ```
> class MyCGSCCPass : public llvm::PassInfoMixin<MyCGSCCPass> {
> public:
>   llvm::PreservedAnalyses run(llvm::LazyCallGraph::SCC &C,
>                               llvm::CGSCCAnalysisManager &AM,
>                               llvm::LazyCallGraph &CG,
>                               llvm::CGSCCUpdateResult &UR) {
>     llvm::FunctionAnalysisManager &FAM =
>         AM.getResult<llvm::FunctionAnalysisManagerCGSCCProxy>(C, CG)
>             .getManager();
>     for (llvm::LazyCallGraph::Node &N : C) {
>       llvm::Function &F = N.getFunction();
>       llvm::PostDominatorTree &PDT =
>           FAM.getResult<llvm::PostDominatorTreeAnalysis>(F);
>       PDT.print(llvm::outs());
>     }
>     return llvm::PreservedAnalyses::none();
>   }
> };
> ```
> 
> Here's a full example of the above:
> https://gist.github.com/modocache/d5804ba567476e32cad1fd0850363532
> 
> I was just reading through this part of the codebase, so I'm a little 
> familiar with it now. Anyway, hope this helps!
> 
> - Brian Gesiak
> 
> On Tue, Jan 7, 2020 at 8:13 AM Mikhail Gudim via llvm-dev 
> <llvm-dev at lists.llvm.org> wrote:
> >
> > Dear all,
> >
> >
> >
> > I would like to use the PostDominatorTree in ArgPromotion. I did not find an example of how to use function level analysis inside CallGraphSCCPass. I tried to follow an example of how to use function-level pass in a module pass, but I hit “llvm_unreachable” in PMDataManager::addLowerLevelRequiredPass.
> >
> >
> >
> > What would be a proper way to make PostDominatorTree available in ArgPromotion?
> >
> >
> >
> > Thanks in advance,
> >
> >
> >
> > Mikhail
> >
> > _______________________________________________
> > LLVM Developers mailing list
> > llvm-dev at lists.llvm.org
> > https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev

-- 

Johannes Doerfert
Researcher

Argonne National Laboratory
Lemont, IL 60439, USA

jdoerfert at anl.gov


More information about the llvm-dev mailing list