[PATCH] Refactor classes in sample profiler for new functionality

[Diego Novillo]

On Tue, Nov 26, 2013 at 2:51 PM, Chandler Carruth <chandlerc at gmail.com>wrote:

>
>   Patch looks fine to commit. I have a question below about code I don't
> understand, but it's not new in this patch I just didn't notice it before.
>
>   I do wonder, would s/Global/Module/g be more clear in an LLVM context?
>

Good point.  Changed.


>
>
> ================
> Comment at: lib/Transforms/Scalar/SampleProfile.cpp:332-333
> @@ -310,9 +331,4 @@
>      Loader.reportParseError("Expected a number, found " + Line);
> -  for (int I = 0; I < NumSymbols; I++) {
> -    StringRef FName = Loader.readLine();
> -    FunctionProfile &FProfile = Profiles[FName];
> -    FProfile.BodySamples.clear();
> -    FProfile.TotalSamples = 0;
> -    FProfile.TotalHeadSamples = 0;
> -  }
> +  for (int I = 0; I < NumSymbols; I++)
> +    Profiles[Loader.readLine()] = SampleFunctionProfile();
>

Well, it is and it isn't.  It's true that we build empty ones later on, but
this one is having the side-effect of reading the symbol table.  I'll get
rid of it when I change the output from the profiler.


Diego.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20131126/01ec5a97/attachment.html>