[PATCH] Enable ARM tail calls for ELF

Eric Christopher echristo at gmail.com
Mon Jul 1 10:37:00 PDT 2013


On Mon, Jul 1, 2013 at 10:28 AM, Quentin Colombet <qcolombet at apple.com> wrote:
> Hi Tim,
>
> The patch LGTM with one question:
> Do you have any performance number to see the impact of this patch?
>
> Indeed, tail calls are generally faster, but I am wondering why it was not
> enabled for non-iOS targets before your patch. Hence, having some evidences
> that it does not have weird side effects would be nice.
>

My guess here is likely testing to be honest, i.e. not wanting to turn
on an untested code path by default.

I would like to see any performance data though. :)

-eric

> Thanks,
>
> -Quentin
>
> On Jun 30, 2013, at 3:12 AM, Tim Northover <t.p.northover at gmail.com> wrote:
>
> Hi,
>
> The attached patch enables tail calls for every target that hasn't
> explicitly been marked as bad (iOS < 5.0). This amounts to any ELF
> target and being more liberal on some ghost MachO targets (e.g.
> arm-apple-darwin9) that no-one probably cares about.
>
> I think the ELF side is justified because the required linker support
> is there in the ARM ELF specification so anywhere it doesn't work has
> a broken linker. I've also tested it on Linux and our MCJIT dynamic
> loader with no issues.
>
> Does anyone object to the change?
>
> Cheers.
>
> Tim.
> <enable-tail.diff>_______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>



More information about the llvm-commits mailing list