[Release-testers] Merging fix for PR26278

[Daniel Sanders via Release-testers]

> -----Original Message-----
> From: hwennborg at google.com [mailto:hwennborg at google.com] On Behalf
> Of Hans Wennborg
> Sent: 03 February 2016 16:49
> To: Daniel Sanders
> Cc: Alexey Samsonov; Release-testers at lists.llvm.org; llvm-
> commits at lists.llvm.org
> Subject: Re: Merging fix for PR26278
> 
> On Wed, Feb 3, 2016 at 1:56 AM, Daniel Sanders
> <Daniel.Sanders at imgtec.com> wrote:
> > It looks like I just missed rc2 but these are now committed on trunk and
> > have gone through the buildbots. Is it ok to merge them to the branch?
> > Together they fix PR26278 and without them Phase 3's 'make check-all' fails
> > on mips without running the majority of the tests.
> >
> > r259493 - [compiler-rt] add_custom_libcxx should pass LLVM_PATH and a
> C++
> > compiler
> >
> > r259542 - Re-commit r259512: [tsan] Add a libc++ and lit testsuite for each
> > ${TSAN_SUPPORTED_ARCH}.
> >
> > r259577 - [tsan] Disable x86_64h build that was enabled as part of r259542.
> >
> > r259593 - [TSan] Use darwin_filter_host_arch to restrict set of test arch on
> > Mac OS.
> 
> Yes, go a head and merge.

Thanks. Merged in r25966[2456].
 
> > Hopefully http://reviews.llvm.org/D16613 will be committed soon which
> will
> > just leave the sanitizer regressions. It looks like I've started to receive
> > patches for the sanitizer regressions but I haven't gone through them yet.
> 
> Sounds good. Keep me posted.

It's looking like these aren't true regressions but are in fact mips64 failures (which weren't run in the 3.7 release). The test names are just a bit misleading.
If this is the case (I've confirmed around half of them) then they won't be release blockers anymore.