[lfs-dev] gcc test failures with current svn

Bruce Dubbs bruce.dubbs at gmail.com
Sun Aug 7 10:20:23 PDT 2016


Ken Moffat wrote:
> I started a build last night on my (slow) Phenom.
>
> In the gcc suite I'm seeing a couple of failures each in g++ and
> libstdc++, whih sounds fine, but the first thing to report is gcc
> itself, which has 52 unexpected failures!   I've checked that the
> ASAN patch was applied - I suspect that without it the failures in
> g++ and libstdc++ would have increased.
>
> I'm using the gold linker.  In binutils, the ordinary tests were ok,
> but in gold 1 of 243 failed:
> FAIL: relro_test.sh
>
>  From gcc:
>
> FAIL: gcc.dg/guality/pr41616-1.c   -O2 -flto -fuse-linker-plugin
> -fno-fat-lto-objects  execution test

Isn't -flto the gold linker?  I'm not sure.

Didn't someone once say that if you don't follow the book and something 
breaks, that you get to keep the pieces?

[snip]
> and then at the end of the reports are a LOT more  (1165) from gcc
> again - those are all in gcc.target/i386/ (I'm building x86_64).
>
> At the moment the build is continuing (in the perl tests, in
> lib/Benchmark, where it has been running for some time).

As you know, I use jhalfs to test changes before I commit to the book. 
These are all the failures I get in gcc:

083-gcc-6.1.0:FAIL: gcc.target/i386/avx2-vpabsb256-2.c execution test
083-gcc-6.1.0:FAIL: gcc.target/i386/avx2-vpabsd256-2.c execution test
083-gcc-6.1.0:FAIL: gcc.target/i386/avx2-vpabsw256-2.c execution test
083-gcc-6.1.0:FAIL: gcc.target/i386/avx2-vpabsb256-2.c execution test
083-gcc-6.1.0:FAIL: gcc.target/i386/avx2-vpabsd256-2.c execution test
083-gcc-6.1.0:FAIL: gcc.target/i386/avx2-vpabsw256-2.c execution test
083-gcc-6.1.0:FAIL: 
experimental/filesystem/iterators/directory_iterator.cc execution test
083-gcc-6.1.0:FAIL: 
experimental/filesystem/iterators/recursive_directory_iterator.cc 
execution tes

The iterator failures are old and the avx2-vpabsb256-2.c failures are from 
a single file that I have not investigated.

I had no failures in binutils.

   -- Bruce



More information about the lfs-dev mailing list