[SOLVED] mpfr_buildopt_tune_case() and MPFR_TUNE_CASE do not match

OstermanA · 01-16-2017, 08:36 PM

I'm trying to compile GCC on a Redhat Enterprise 6 box, and mpfr 3.1.5 is failing this part of make check:

Code:

======================================
   MPFR 3.1.5: tests/test-suite.log
======================================

# TOTAL: 160
# PASS:  158
# SKIP:  1
# XFAIL: 0
# FAIL:  1
# XPASS: 0
# ERROR: 0

.. contents:: :depth: 2

FAIL: tversion
==============

[tversion] MPFR 3.1.5
[tversion] Compiler: GCC 6.3.0
[tversion] GMP: header 6.1.2, library 6.1.2
[tversion] TLS = yes, decimal = no, GMP internals = no
[tversion] intmax_t = yes, printf = yes
[tversion] gmp_printf: hhd = yes, lld = yes, jd = yes, td = yes, Ld = yes
ERROR! mpfr_buildopt_tune_case() and MPFR_TUNE_CASE do not match!
  src/x86_64/core2/mparam.h
  default
FAIL tversion (exit status: 1)

SKIP: tget_set_d64
==================

SKIP tget_set_d64 (exit status: 77)

I tried the same process on another machine, and it worked fine. Tracked down the output from the test on that machine and it's coming back with 'default'. Base on what tversion.c says, it looks like mpfr_buildopt_tune_case() is somehow coming back with the Core2/Xeon response, while the compiler directives are set to 'default'.

/proc/cpuinfo shows that the machine is a "Intel(R) Xeon(R) CPU E5-4640 v2 @ 2.20GHz".

I do not know what to do next. Any suggestions?

hydrurga · 01-18-2017, 07:41 AM

Did you run make tune beforehand, as described here?: http://lists.gforge.inria.fr/piperma...er/006650.html

OstermanA · 01-18-2017, 11:48 PM

As if by magic, my build script no longer fails on make check, and make tune shows "nothing to do". I do not understand what changed. Either the phase of the moon changed enough that it works again, or the system admins changed something without telling us again.

hydrurga · 01-19-2017, 01:20 AM

Quote:

Originally Posted by OstermanA

As if by magic, my build script no longer fails on make check, and make tune shows "nothing to do". I do not understand what changed. Either the phase of the moon changed enough that it works again, or the system admins changed something without telling us again.

:-) The bountifulness of the Computer Gods is limitless.