[llvm] r209089 - ARM: make libcall setup more table driven

[Saleem Abdulrasool]

On Sun, May 18, 2014 at 11:32 AM, Renato Golin <renato.golin at linaro.org>wrote:

> On 18 May 2014 17:39, Saleem Abdulrasool <compnerd at compnerd.org> wrote:
> > Author: compnerd
> > Date: Sun May 18 11:39:11 2014
> > New Revision: 209089
> >
> > URL: http://llvm.org/viewvc/llvm-project?rev=209089&view=rev
> > Log:
> > ARM: make libcall setup more table driven
> >
> > Rather than create a series of function calls to setup the library
> calls, create
> > a table with the information and just use the table to drive the
> configuration
> > of the library calls.  This makes it easier to both inspect the list as
> well as
> > to modify it.  NFC.
>
> Hi Saleem,
>
> I agree this is an improvement, but there are other places that set
> libcalls names, not to mention other targets. I think such a change
> should have been agreed on the list by other targets' maintainers. At
> the very least, move the struct up the hierarchy with a
> table-driven-setter, like we do on the loop vectorizer's cost-table.
>

Yes, I agree that there are other targets and places that set libcalls.
 ARM and Hexagon are both particularly large in the set of library calls
that they initialize.  Im not sure how many of the other targets would
really benefit from such a generic infrastructure.


> Can you start the conversation going on the list? No need to revert
> the patch, but would be good to make sure this is a design we all
> follow.
>

Sure, this sounds like a good idea.  I think that getting the opinion of
others on whether this is enough of a pain on other targets to warrant a
generic infrastructure for this is worthwhile.


> cheers,
> --renato
>

-- 
Saleem Abdulrasool
compnerd (at) compnerd (dot) org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140518/96c1c576/attachment.html>