[PATCH] D13259: LLE 6/6: Add LoopLoadElimination pass

Hal Finkel via llvm-commits llvm-commits at lists.llvm.org
Fri Oct 30 10:00:55 PDT 2015


----- Original Message -----
> From: "Daniel Berlin" <dberlin at dberlin.org>
> To: reviews+D13259+public+975ad6db6ce955f7 at reviews.llvm.org
> Cc: anemet at apple.com, "Hal Finkel" <hfinkel at anl.gov>, mssimpso at codeaurora.org, "Sanjoy Das"
> <sanjoy at playingwithpointers.com>, "llvm-commits" <llvm-commits at lists.llvm.org>
> Sent: Friday, October 30, 2015 11:07:45 AM
> Subject: Re: [PATCH] D13259: LLE 6/6: Add LoopLoadElimination pass
> 
> I'm confused.
> AFAIK, we normally do *not* prefix the messages explicitly (i see
> this pass does it in some places but not others, that should be
> changed to be not prefixed, IMHO).
> 
> 
> This pass has DEBUG_TYPE set to loop-load-elim at the top, so if you
> use debug-only, you won't get them unless you specify
> debug-only=loop-load-elim.
> 
> 
> I would suggest rather than prefix *anything*, anywhere, that if we
> want prefixes, we just have the DEBUG macro output the debug type in
> front of the message ;-)
> 
> Then it's automatic, not confusing, and clear where it came from.
> 

+1

 -Hal

> 
> On Fri, Oct 30, 2015 at 9:01 AM, Renato Golin <
> renato.golin at linaro.org > wrote:
> 
> 
> rengolin added a comment.
> 
> While working on top of this patch, I noticed that not all DEBUG()
> messages are prefixed with "LLE:", which may confuse people using
> -debug-only=.
> 
> 
> http://reviews.llvm.org/D13259
> 
> 
> 
> 
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory


More information about the llvm-commits mailing list