[lld] r249752 - Revert: r249728 - Roll back r249726 and r249723 because they broke buildbots.

[Hal Finkel via llvm-commits]

----- Original Message -----
> From: "Rafael Espíndola" <rafael.espindola at gmail.com>
> To: "Shankar Easwaran" <shankare at codeaurora.org>
> Cc: "Hal Finkel" <hfinkel at anl.gov>, "llvm-commits" <llvm-commits at lists.llvm.org>
> Sent: Tuesday, October 13, 2015 12:41:31 PM
> Subject: Re: [lld] r249752 - Revert: r249728 - Roll back r249726 and r249723 because they broke buildbots.
> 
> On 13 October 2015 at 13:38,  <shankare at codeaurora.org> wrote:
> > I think the start address of the image, is ABI defined isn't it ? A
> > target
> > specific way of setting the start address would be necessary IMO.
> 
> *If* that is the case, please add a reference to the relevant ABI
> document.
> 
> If not, please revert the getVAStart part of the change.

Right. There are two issues here:

 1. Can we always used a fixed VAStart address?

 2. Should it be a target property?

The answer, at least in the long run, is certainly no. There are plenty of applications that run under embedded (or other light-weight) kernels that need the ability to specify a starting virtual address in order to function correctly (or, at least, to not waste huge amounts of memory).

However, regarding (2), this seems much less clear. It might be more natural for this to be a property of the target operating system more than the target, for example. But, in any case, lacking specific documentation and/or understanding on how to choose the values, I'll revert this.

 -Hal

> 
> Cheers,
> Rafael
> 

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