release_35 branch patches

[Hal Finkel]

----- Original Message -----
> From: "Brad Smith" <brad at comstyle.com>
> To: "Hal Finkel" <hfinkel at anl.gov>
> Cc: llvm-commits at cs.uiuc.edu, "Bill Wendling" <isanbard at gmail.com>
> Sent: Sunday, July 27, 2014 10:04:32 PM
> Subject: Re: release_35 branch patches
> 
> On 27/07/14 11:01 PM, Hal Finkel wrote:
> > ----- Original Message -----
> >> From: "Brad Smith" <brad at comstyle.com>
> >> To: llvm-commits at cs.uiuc.edu
> >> Sent: Sunday, July 27, 2014 9:53:48 PM
> >> Subject: release_35 branch patches
> >>
> >> Hi,
> >>
> >> I'd like to propose a few patches to be back ported to the 3.5
> >> branch..
> >>
> >> r213966
> >> Fix arc4random detection.
> >>
> >> r213899
> >> Don't use 128bit functions on PPC32.
> >
> > I approve this one.
> >
> >>
> >> r213960
> >> [PowerPC] Support TLS on PPC32/ELF
> >
> > This touches common code, is a new feature, and thus we decided not
> > to do this.
> 
> What? This close to the release and it is left out? That's beyond
> disappointing and illogical when the PIC diff went in which also
> touches "common code".

The patch was not ready until almost a week after 3.5 branched. The PIC diff was ready earlier. Please don't be insulting; several of us working on the PowerPC backend discussed this on IRC, with Justin (the author of the patch), and other members of the community. To be clear, my understanding is that even with the TLS patch applied, we cannot self-host on ppc32/ELF. Given the continuing efforts of Justin and others over the next few months, I'm looking forward to Clang/LLVM 3.6 being a viable option for ppc32/ELF.

 -Hal

> 
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
> 
> 

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